Differential Receptor Subunit Affinities of Type I …...Type I interferons (IFNs) elicit antiviral,...

doi:10.1016/j.jmb.2006.11.053 J. Mol. Biol. (2007) 366, 525–539

Differential Receptor Subunit Affinities of Type IInterferons Govern Differential Signal Activation

Eva Jaks1, Martynas Gavutis1, Gilles Uzé2, Jacques Martal3

and Jacob Piehler1⁎
1Institute of Biochemistry,Johann WolfgangGoethe-UniversityFrankfurt am Main,Germany2UMR 5124, CNRS,Montpellier, France3Station de Physiologieanimale, INRA Jouy-en-Josas,France
Abbreviations used: IFN, type I inifnar2, subunits of the human type IEC, extracellular domain; RIfS, reflespectroscopy; TIRFS, total internal rspectroscopy; OG488, Oregon GreenHepes-buffered saline; ISGF3, interffactor 3.E-mail address of the correspondi

[email protected]

0022-2836/$ - see front matter © 2006 E

Type I interferons (IFNs) elicit antiviral, antiproliferative and immunmo-dulatory responses by binding to a shared cell surface receptor comprisingthe transmembrane proteins ifnar1 and ifnar2. Activation of differentialresponse patterns by IFNs has been observed, suggesting that members ofthe family play different roles in innate immunity. The molecular basis fordifferential signaling has not been identified yet. Here, we have investigatedthe recognition of various IFNs including several human IFNα species,human IFNω and human IFNβ as well as ovine IFNτ2 by the receptorsubunits in detail. Binding to the extracellular domains of ifnar1 (ifnar1-EC)and ifnar2 (ifnar2-EC) was monitored in real time by reflectance interferenceand total internal reflection fluorescence spectroscopy. For all IFNsinvestigated, competitive 1:1 interaction not only with ifnar2-EC but alsowith ifnar1-EC was shown. Furthermore, ternary complex formation wasstudied with ifnar1-EC and ifnar2-EC tethered onto solid-supportedmembranes. These analyses confirmed that the signaling complexesrecruited by IFNs have very similar architectures. However, differences inrate and affinity constants over several orders of magnitude were observedfor both the interactions with ifnar1-EC and ifnar2-EC. These data werecorrelated with the potencies of ISGF3 activation, antiviral and anti-proliferative activity on 2fTGH cells. The ISGF3 formation and antiviralactivity correlated very well with the binding affinity towards ifnar2. Incontrast, the affinity towards ifnar1 played a key role for antiproliferativeactivity. A striking correlation was observed for relative binding affinitiestowards ifnar1 and ifnar2 with the differential antiproliferative potency.This correlation was confirmed by systematically engineering IFNα2mutants with very high differential antiproliferative potency.

© 2006 Elsevier Ltd. All rights reserved.

Keywords: type I interferon receptor; protein–protein interaction; solid-supported membrane; interferon-beta; interferon-alpha
*Corresponding author
Introduction

Type I interferon (IFN) signaling plays a key rolein the innate immune response against viral andbacterial infection as well as malignancies.1 Binding

terferon; ifnar1,interferon receptor;

ctometric interferenceeflection fluorescence488; HBS,

eron-stimulated gene

ng author:

lsevier Ltd. All rights reserve

of IFNs to their shared cell surface receptor ifnar1and ifnar2 subunits activates numerous signaltransduction cascades resulting in a pleiotrophiccellular response.2 The cellular response upontreatment with IFNs is typically discriminated intoantiviral, antiproliferative and immunmodulatoryactivities. In comparison to other cytokine receptorsthe high redundancy of ligands (∼15 IFNα sub-types, one IFNβ, one IFNκ, one IFNω in humans)3

binding to the same receptor is unusual, suggestingdifferential functional properties. Indeed, numerousinstances of apparently differential cellular responseupon treatment with different IFNs have beenreported.4–9 IFNs are already successfully used forthe treatment of rather different diseases,10 andunderstanding the molecular basis of such differ-

d.

mailto:[email protected]

http://dx.doi.org/10.1016/j.jmb.2006.11.053

526 Differential Type I Interferon Receptor Affinities

ential signaling could tremendously contributetowards systematical engineering of IFNs for med-ical application. However, the assessment of differ-ential signal activation is extremely difficult forseveral reasons: different cell types have been usedin different studies and for different activity assays,and differential responses appear to be dependenton the cellular context. Furthermore differentstrategies have been used for defining differentialresponses. In many cases, antiviral protection andthe inhibition of cell proliferation were compared fordifferent IFNs on the level of dose-response curves.While similar, relatively low IFN concentrations arerequired for antiviral protection, more different, andin some cases also much higher IFN concentrationsare required for eliciting antiproliferative res-ponse.11 Only a few studies are available wherethe cellular response was quantified on the mRNAlevel.12–14 In these studies differential gene activa-tion was observed at low, but not at high IFN doses,confirming the key role of IFN concentration formodulating the cellular response. All these datataken together suggest that non-linear dose-response and affinity-response behavior, which hasalso been observed for other cytokine receptors,15

may play a role for differential signal activation.The interaction of IFNα2 and IFNβ with the high-

affinity subunit ifnar2 has been investigated in vitrousing the extracellular domain (ifnar2-EC), andextended structure-function studies have been car-ried out.16–21 While only minor differences wereobserved in the structure of the complexes of ifnar2-EC with IFNα2 and IFNβ,22 a ten to 20-fold higheraffinity of IFNβ for ifnar2-EC was observed. Whilethe interaction of IFNs with the extracellular domainof ifnar1 (ifnar1-EC) appears rather complex,23,24

competitive binding and largely overlapping bind-ing epitopes were observed for IFNα2 and IFNβ.25

These binding studies suggested that IFNα2 andIFNβ recruit a 1:1:1 complex with ifnar2 and ifnar1,which probably have rather similar architectures.25

However, a 100-fold higher affinity towards ifnar1-EC was obtained for IFNβ compared to IFNα2.26

Analysis of ligand binding to ifnar1-EC and ifnar2-EC on solid-supported lipid bilayers revealeddynamic equilibria between binary and ternarycomplexes,26,27 which are determined by the rateconstants of the interaction with the receptorsubunits in plane of the membrane. These resultssuggested that the ternary complex stability and theefficiency of recruitment of ifnar1 into the ternarycomplex may play a key role for differentialsignaling and responsiveness. Indeed, the interac-tion with ifnar1 was shown to be important forabsolute and relative antiproliferative activity.20,28,29

Here, we have scrutinized a set of different IFNswith respect to receptor subunit binding and signalactivation, namely human IFNα1, IFNα2, IFNα8,IFNα21, IFNβ, and IFNω, as well as ovine IFNτ2(trophoblastin). Trophoblastins are only found inruminants, but also activate the human type Iinterferon receptor,30 and have been reported toexert particularly low antiproliferative acivities.31,32

Kinetic rate constants and equilibrium dissociationconstants were determined by solid phase bindingassays with the receptor subunits ifnar1-EC andifnar2-EC in vitro. Furthermore, assembly of theternary complex was studied on solid-supportedmembranes. Correlation with cellular signal activa-tion and responses confirmed the key role of absoluteaffinity towards ifnar1 for antiproliferative activity.More importantly, however, the relative affinity tothe receptor subunits was shown to play a key rolefor the differential activity of type I IFNs. Based onthese insights,we successfully engineered type I IFNswith highly differential activity of IFNα2 mutants.

Results

IFNs have very different affinities towards thereceptor subunits

The interaction of the different IFNs with ifnar2-EC and ifnar1-EC was studied by real-time solidphase detection in order to identify differences inaffinity and rate constants. The receptor subunitswere site-specifically immobilized through their C-terminal decahistidine-tags onto silica surface of thetransducer using multivalent chelator heads.33 Non-specific binding of IFNs was negligible at thesesurfaces after blocking excess immobilization siteswith MBP-H10. Binding of IFNs was monitored in alabel-free manner by RIfS. Typical ligand bindingexperiments are shown in Figure 1(a). Fast associa-tion kinetics were observed for most IFN subtypes(Figures 1(a) and 2(a)), while strong differences indissociation kinetics were found (Figures 1(a) and(b) and 2(a) and (b)). For IFNτ2, much fasterdissociation was observed than for the humanIFNs. For the fast dissociating IFNα1 and IFNτ2,the equilibrium dissociation constant were deter-mined by determining equilibrium binding atdifferent ligand concentrations (Figure 1(c) and(d)). From these titrations, a KD of ∼100 nM wasobtained for IFNα1, and a KD of 1200 nM for IFNτ2.The interaction of fluorescence-labelled IFN with

immobilized ifnar2-EC was also monitored by totalinternal reflection fluorescence spectroscopy(TIRFS). For this purpose, IFNs were site-specificallylabelled at free cysteine residues (S136C/S137Cmutants) with Oregon Green 488 (OG488) malei-mide (referred to as OG488IFN in the following). Verysimilar binding rates were obtained for the labelledIFNs compared to the unlabelled (data not shown).In order to suppress rebinding, unlabelled IFN wasadded during the dissociation phase (Figure 1(f)).Thus, dissociation rate constants unbiased byrebinding were determined for the IFN subtypes,which are summarized in Figure 1(g). Also, theinteraction kinetics of fluorescence-labelled IFNτ2could be resolved due to much better signal-to-noiseratio compared to label-free detection.27 A kd of0.7(±0.2) s−1 and a ka of (5±3)×105 M−1s−1 wereobtained from these binding assays, yielding a KD of

Figure 1. Binding of different IFNs to immobilized ifnar2-EC. (a) Binding curves for 100 nM of IFNα2 (black) and100 nM IFNα1 (green). (b) Comparison of the dissociation curves for IFNβ (red), IFNω (blue), IFNα2 (black), IFNα1(green) and IFNτ2 (orange). (c) Binding of IFNα1 at different concentrations to immobilized ifnar2. (d) Comparison of theequilibrium responses for IFNα2 and IFNτ2 at different concentrations. The equilibrium constants were determined byfitting the law of mass action. (e) Dissociation kinetics of IFNτ2 as monitored by TIRFS with site-specifically labelledIFNτ2. (f) Rebinding-biased dissociation of fluorescence-labelled IFNα21 (black) was assessed by chasing with 1 μMunlabelled IFNα21. (g) Comparison of the dissociation rate constants. (h) Comparison of the equilibrium dissociationconstants.

527Differential Type I Interferon Receptor Affinities


1.4 μM, which is in good agreement with the KDobtained from equilibrium titration with the unla-belled IFNτ2. Thus, the association rate constant ofIFNτ2 is approximately one order of magnitudelower than the association rate constants of thehuman IFN investigated here. The average KDvalues obtained for different IFNs are compared inFigure 1(h), yielding differences over three orders ofmagnitude between different IFNs.The interaction of the IFNs with ifnar1-EC was

probed in a similar manner. Micromolar proteinconcentrations were required in order to detectbinding for IFNα subtypes as well as IFNτ2, andvery transient binding was observed. Typical bind-ing curves are shown in Figure 2(a). For IFNω andIFNβ, substantially higher binding affinities wereobserved, and the ligand binding kinetics could bewell resolved (Figure 2(b)). The binding affinities of

Figure 2. Binding of different IFNs to immobilized ifnar1-IFNω (blue). (b) Comparison of the dissociation curves for IFNBinding signal for IFNα1 injected at different concentrationIFNα1 (red) and IFNτ2 (black). (e) Comparison of the eqdissociation rate constants.

the IFNα subtypes and IFNτ2 were determinedfrom the equilibrium binding signals (Figure 2(c)and (d)), and are shown in comparison in Figure2(e). These KD values were all in the μM range, butIFNα1, IFNα8 and IFNα21 had significantly higheraffinity towards ifnar1-EC than IFNα2, while IFNτ2had significantly lower affinity. For IFNβ and IFNω,the KD was determined from the binding kinetics.For both IFNω and IFNβ, similar association rateconstants of ∼2×105 M−1s−1 were obtained. Therate constants for the IFNα subtypes and IFNτ2were determined by TIRFS. For this purpose, theIFNα subtypes were site-specifically fluorescence-labelled by genetically incorporating an additionalcysteine (S136C for IFNα2 and S137C for the otherIFNα subtypes). The relative dissociation rateconstants were in good agreement with the relativeequilibrium dissociation constants (Figure 2(f)).

EC. (a) Binding curves for 10 μM IFNα2 (black) and 1 μMβ (red), IFNω (blue), IFNα2 (black) and IFNα1 (green). (c)s. (d) Concentration-dependent equilibrium response foruilibrium dissociation constants. (f) Comparison of the


Thus, very similar association rate constants of∼2×105 M−1s−1 were observed for all IFNs for theinteraction with ifnar1-EC.

All IFNs bind competitively to both ifnar2 andifnar1

While it is well established that all type I IFNsbind competitively to the same cell surface receptor,differential binding to the receptor subunits havefrequently been proposed to be responsible fordifferential signal activation. In order to studycompetition of different IFNs, we simultaneouslyinjected a fluorescence-labelled and a non-labelledIFN and monitored binding by RIf and TIRFS.Typical competition experiments with ifnar2-EC andifnar1-EC are shown in Figure 3. In the beginning,ligand binding is kinetically controlled and thereforesimilarly fast for most IFNs, and fast increase of thefluorescence signal is observed in case of an excess oflabelled ligand. But if the kd of the labelled ligand ishigher than the kd of the unlabelled ligand, thelabelled ligand will be exchanged by the unlabelledduring the injection. Thus, competition of labelledligand and unlabelled ligand was directly revealedby a decay of the fluorescence signal during theinjection period. For all combinations of IFN,competition for the same binding site was observednot only for the binding to ifnar2-EC (Figure 3(a)),but also for the binding to ifnar1-EC (Figure 3(b)),which has been speculated to differentially engageIFNs.16,34 Furthermore, an ifnar1-EC sub-fragmentcontaining the three N-terminal Ig-like domains,which was previously shown to be required andsufficient for ligand recognition,25 was also suffi-cient for binding for all subtypes.

Ternary complex formation on supported lipidbilayers

Cross-linking of the receptor subunits by simul-taneous interaction with the ligand is believed to

Figure 3. Competition binding of IFNs to the ifnar subuifnar2-H10 in the presence of unlabelled IFNα2 in different concurves were normalized to the maximum fluorescence signal.H10 in the absence of competitor (black) and with 100 nM (reperiod.

induce signaling by type I IFNs. We have recentlypresented an approach to mimic ternary complexformation on the plasma membrane by tetheringifnar1-EC and ifnar2-EC onto solid-supported lipidbilayers.21 Here, we studied ternary complexformation with the different IFNs by label-freedetection in order to identify potential differencesin complex stoichiometries. A typical bindingexperiment with stoichiometric amounts of ifnar2-EC and ifnar1-EC on solid-supported lipid bilayersis shown in Figure 4(a). Upon injection of IFNα21formation of a stable 1:1:1 complex was observed,and dissociation of the ligand was hardly detect-able. Based on this assay, the stoichiometry of theternary complex was confirmed to be 1:1:1 for allIFNs. The ligand dissociation kinetics was depen-dent on the surface concentration of the receptorsubunits, which was studied in detail by simulta-neous TIRFS-RIf detection27 using fluorescencelabelled IFNs. Dissociation of IFNα21 at differentsurface concentrations of ifnar1-EC and ifnar2-ECin stoichiometric ratio is shown in Figure 4(b). Withincreasing surface concentrations slower dissocia-tion was observed. These dissociation curves werefitted by a two-step dissociation model as describedpreviously.27 A similar effect was observed whenthe concentration of only one of the receptorsubunits was increased as shown in Figure 4(c)for IFNτ2. A comparison of the dissociationkinetics for different IFNs at the same (stoichio-metric) surface concentration of ifnar1-EC andifnar2-EC is shown in Figure 4(d). The observeddissociation kinetics were in good agreement witha two-step dissociation model parameterized withthe binding constants obtained from the interactionstudies with the individual receptor subunits.Taken together, these results confirmed that thestoichiometry and the architecture of the signalingcomplex recruited by different IFNs are verysimilar, suggesting that differential affinity towardsthe receptor subunits may be the reason fordifferential signal activation.

nits. (a) Binding of 100 nM OG488IFNα21 to immobilizedcentrations (red, 50 nM; green, 100 nM; blue, 200 nM). The(b) Binding of 200 nM OG488IFNα21 to immobilized ifnar1-d) and 200 nM (green) IFNβ. The bar marks the injection

Figure 4. Interaction of IFNs with ifnar2-EC and ifnar1-EC tethered onto supported lipid bilayers. (a) Typical bindingassay detected by RIfS including tethering of ifnar2-EC (I) and ifnar1-EC (II), as well as injection of IFNα21 (III). (b)Comparison of the dissociation of OG488IFNα21 from ifnar1-EC and ifnar2-EC tethered to the membrane in stoichiometricratio at different surface concentrations (blue, 40 fmol/mm2; red, 8 fmol/mm2; black, ifnar2 only). (c) Dissociation ofOG488IFNτ2 at constant surface concentration of ifnar2-EC and increasing surface concentrations of ifnar1-EC(concentrations are indicated in the inset). (d) Comparison of the dissociation of different fluorescence-labelled IFNs atlow receptor surface concentrations (2 fmol/mm2) as detected by TIRFS (orange, IFNβ; red, IFNα2; blue, IFNα21; green,IFNα1; black, IFNτ2).


Cellular activities

For characterizing differential activities of the IFNused here, we determined some cellular activities ofthe IFN by standard assays using 2fTGH andHL116, two clones derived from the HT1080 cellline, for all assays. Fast responses such as theformation of the transcription factor ISGF3 andantiviral protection were determined after 6 h ofligand incubation. Antiproliferative activity wasdetermined after ligand incubation for seven days.Since the medium was not changed during theassay, autocrine signaling may also be involvedunder these condition, as previously shown for celldifferentiation assays.9 Typical dose-responsecurves for ISGF3 formation induced by differentIFNs are shown in Figure 5(a), and in Figure 5(b) forthe antiproliferative activities. The EC50 valuesobtained from these curves are compared in Figure5(c) and (d). Very different EC50 values wereobtained in the ISGF3 formation assay, rangingfrom ∼2 pM for IFNβ to ∼4 nM for IFNτ2. Alsoamong the IFNα subtypes, substantial differenceswere observed, with IFNα8 being the most potent,and IFNα1 being nearly 100-fold less potent thanIFNα8. ISGF3 formation has been reported to

correlate well with antiviral activity, which wasconfirmed for the different IFNs by antiviral protec-tion assays against infection by vesicular stomatitisvirus (VSV). The EC50 values required for antiviralprotection were ∼tenfold higher than for the ISGF3activation (Figure 5(c) and (e)), but a very similarpattern was observed (Figure 5(f)). A rather differentpattern was obtained for the antiproliferativeactivities. Here, very similar antiproliferative activ-ities were observed for the IFNα subtypes, whileIFNω and in particular IFNβ were substantiallymore potent. In general, much higher IFN concen-trations were required for inducing an antiprolifera-tive response compared to ISGF3 formation andantiviral protection. Thus, nearly micromolar con-centrations of IFNτ2 were required for obtainingantiproliferative activity.

Differential activities correlate with relativebinding affinities

Strikingly, the affinity pattern observed for theinteraction of different IFNs with ifnar2-EC (Figure1(h)) resembles the activity patterns observed in theISGF3 formation assays (Figure 5(c)), while theaffinity pattern for the interaction with ifnar1-EC

Figure 5. Cellular activities of different type I IFNs. (a) Concentration-dependent ISGF3 activation by different IFNs asmeasured by reporter luciferase activity in HL116 cells. Concentration-dependent antiproliferative activity by differentIFNs on 2fTGH cells. (c)–(e) Comparison of the EC50 values obtained for different IFNs for ISGF3 activation (c),antiproliferative activity (d) and antiviral activity (e). (d) Correlation of the IC50 values for ISGF3 activation and antiviralactivity.


(Figure 2(e)) resembles the antiproliferative activitypattern (Figure 5(d)). For correlating ligand affinitytowards the receptor subunits with the cellularactivities, we converted the EC50 values intopotencies relative to the potency of IFNα2 (c.f.Methods). A direct comparison of these patternsnormalized to the respective affinities and activitiesof IFNα2 are shown in Figure 6(a) and (b),respectively. The relative ISGF3 formation activityindeed correlates very well with the binding affinity

towards ifnar2 (Figure 6(a)). Only for IFNβ, sub-stantially higher affinity compared to activity wasobserved. It is in principle possible that at theseextremely low concentrations the apparent activityof IFNβ, which is known to stick to surfaces, isdecreased by loss of material during the assay.However, the fact that affinity surpasses activity hasalso been observed for other cytokine receptors.15

It has been suggested that the affinity towardsifnar1 plays a key role for differential signaling by

Figure 6. Correlation between affinities and activities. (a) Normalized potency in the ISGF3 formation assayscompared to the normalized affinity towards ifnar2-EC. (b) Normalized antiproliferative potency compared to thenormalized affinity towards ifnar1-EC. (c) Antiproliferative potency relative to the ISGF3 potency compared to thenormalized affinity towards ifnar1-EC. (d) Antiproliferative potency relative to the ISGF3 formation potency compared tothe square root of the affinity towards ifnar1-EC relative to the affinity towards ifnar2-EC.


IFNs, and the relatively high antiproliferativeactivity of IFNβ is one of the most prominentexamples.29 Strikingly, relatively good correlationbetween the relative antiproliferative activities andthe affinity towards ifnar1 was observed, yet withsome substantial deviations, which we attributed tothe different binding to the ifnar2. We thereforecalculated the antiproliferative potency relative tothe ISGF3 formation potency as a measure fordifferential activity (termed differential antiproli-ferative potency, in the following), which wascompared with the relative affinities towardsifnar1-EC (Figure 6(c)). While in principle similarpatterns were obtained, large deviations wereobserved, which systematically correlated with theaffinity towards ifnar2: for IFNβ and IFNω withhigh affinities towards ifnar2, the relative affinitytowards ifnar1 was much higher than the differ-ential antiproliferative potency; for IFNα1 andIFNτ2 with low affinities towards ifnar2, thedifferential antiproliferative potency was muchhigher than the relative ifnar1 affinity. This observa-tion indicated that the relative affinities towardsifnar1 and ifnar2 are responsible for differentialantiproliferative potency. For this reason, we usedthe ratio of the binding affinities towards ifnar1 andifnar2 for the correlation with the differentialantiproliferative potency. The best correlation was

observed for the square root of the relative affinities(Figure 6(d)). This correlation is in line with thestrong differential antiproliferative potency ofIFNα1, which shows only a moderate increase inthe affinity towards ifnar1, but a substantialdecrease in affinity towards ifnar2.

Engineering IFNα2 variants with differentialactivities

Based on the correlation between the relativeaffinities towards the receptor subunits and thedifferential antiproliferative potency we aimed toengineer IFNα2 mutants with differential activities.In order to mimic the activity pattern observed forIFNα1, we decreased the affinity towards ifnar2 bythe mutation R144A,19 and increased the affinitytowards ifnar1 by the mutation E58A.35 The relativeaffinities towards ifnar1 and ifnar2 for the indivi-dual and combined mutations determined by ligandbinding assays (cf. Table 1) are compared with therelative pISGF3 and the differential pAP in Figure 7(a)and (b), respectively. Similar correlations as for thedifferent wild-type IFNs were obtained for theIFNα2 mutants. Thus, the relative pISGF3 patternvery closely followed the pattern of the relativeaffinity towards ifnar2 (Figure 7(a)). The differentialpAP pattern followed the affinity towards ifnar1 with

Table 1. Interaction constants and activities determined for different IFNs and IFNα2 mutants

IFN

Ifnar2-EC Ifnar1-EC Activities (EC50)

ka (M−1s−1) kd (s−1) Kd (nM) ka (M

−1s−1) kd (s−1) Kd (μM) ISGF3 (pM) Antiproliferative (nM) Antiviral (pM)

IFNα1 2×106 0.12 100 ∼2×105 0.5 2.5 47 2 400IFNα2 3×106 0.015 5 ∼2×105 1 5.0 8 1.1 40IFNα8 6×106 0.02 3 ∼2×105 0.5 2.2 5 1.1 20IFNα21 3×106 0.08 25 ∼2×105 0.5 2.5 20 1.5 50IFNβ 1×107 0.001 ∼0.1 ∼2×105 0.025 0.05 1.5 0.065 –IFNω 6×106 0.008 1 ∼2×105 0.08 0.40 3 0.2 –IFNτ2 5×105 0.8 1200 ∼2×105 2 10.0 3800 60 4000E58A 3×106 0.015 5 ∼2×105 0.1 0.5 2 0.14 –R144A 5×105 0.05 150 ∼2×105 1 5.0 575 10.0 –E58A, R144A 5×105 0.05 150 ∼2×105 0.1 0.5 390 3.2 –Error (%) 20 10 20 40 30 20 30 30


substantial deviations for the mutants R144A andE58A (Figure 7(b)). Strikingly, the differential pAP

increased not only upon increasing the affinitytowards ifnar1 (E58A-mutation), but even substan-tially more upon decreasing the affinity towardsifnar2 (R144A-mutation). For the combined muta-tion IFNα2-ER, the highest differential antiprolifera-tive potency was observed, even higher than forIFNα1. While the ISGF3 activity of this mutant is 50-fold lower than that of wild-type IFNα2, onlythreefold lower antiproliferative activity wasobserved (cf. Table 1 and Figure 7(d)). The correla-tion of the differential antiproliferative potency with

Figure 7. Correlation between affinities and activities. (a)ifnar1-EC. (b) Relative ISGF3 activity and relative differeantiproliferative potency and square root of the relative affinitidose-response curves for ISGF3 activation (circles) and antiprothe mutant IFNα2-ER (black).

the square root of the relative affinities was also wellconfirmed for all these mutants. These resultscorroborate, that the surprisingly high antiprolifera-tive response of IFNα1 is probably due to its moresymmetric affinities towards ifnar1 and ifnar2.

Discussion

Induction of differential response patterns bydifferent type I IFNs has been frequently described,but the molecular and biophysical basis of howdifferential signaling through the same cell surface

Relative affinities towards ifnar2-EC (blue) and towardsntial antiproliferative potency. (c) Relative differentiales towards ifnar1-EC and ifnar2-EC. (d) Comparison of theliferative activity (diamonds) of IFNα2 wild-type (red) and


receptor is achieved has remained enigmatic. Manyspeculations such as different structures of theligand-receptor complexes, different orientations ofthe complexes, different complex stoichiometriesand complex compositions, or even additionalcomponents have been suggested to be responsiblefor this phenomenon. Here, we have studied theinteractions of a representative selection of the type IIFN family with the extracellular domains of thereceptor subunits ifnar1-EC and ifnar2-EC. For allthese IFNs, exclusively 1:1 complexes were formedwith the individual subunits, and complex stoichio-metries of 1:1:1 were observed for the ternarycomplexes formed on supported lipid bilayers.Furthermore, no indication of homodimeric ligand-receptor complexes were detectable. The mostnotable differences between the IFNs were thedifferent affinities and kinetics of the interactionwith the receptor subunits. Compared to the alreadywell characterized IFNα2, we observed faster dis-sociation from ifnar2-EC for IFNα8, IFNα21 (inparticular for IFNα1) while the association kineticsof these IFNs were similar to the very fast associa-tion kinetics of IFNα2. For IFNω and IFNβ,significantly higher affinities towards ifnar2-ECwere obtained, which could only be estimatedbecause of the fast, diffusion-controlled associationkinetics. In contrast, IFNτ2 dissociates from ifnar2

Figure 8. Electrostatic properties of different IFNs at theirto a homologue residue in the center of the binding site as deterand IFNτ R150 in (a), and IFNα2 Y89, IFNβ Y92 and IFNτ Q9SwissPdbViewer.44

even faster than IFNα1, but also associates substan-tially slower than all other IFNs, resulting in a∼300-fold lower affinity than IFNα2. Thus, differentaffinities towards ifnar2-EC were based on largevariations of association kinetics as well as dissocia-tion kinetics. It was shown before that the associa-tion of IFNα2 is enhanced by electrostatic steering,which is based on positively charged amino acidresidues in the binding site (Figure 8(a)) and itsvicinity being attracted by the strongly negativeelectrostatic potential of ifnar2-EC.36 The even fasterassociation of IFNβ and IFNω can be explained bytheir even stronger positively charged surfaces(Figure 8(a)). In contrast, the binding site of IFNτ2is rather negatively charged, which explains themuch slower association kinetics compared to thehuman IFNs.The IFNs also exhibited very different affinities

towards ifnar1, which was in all cases at least oneorder of magnitude lower than towards ifnar2. Yet,some distinct differences in the affinity pattern wereobserved, which may be critical for differentialsignaling. All IFNα subtypes showed similarly lowaffinities towards ifnar1, but the affinities of IFNα1,IFNα8 and IFNα21 were still twofold higher thanthe affinity of IFNα2. Both IFNω and IFNβ exhibitedsubstantially higher affinities than IFNα2, andIFNτ2 slightly lower. In contrast to the interaction

binding sites for ifnar2 (a) and ifnar1 (b). The arrows pointmined for IFNα219,35 and IFNβ20 (IFNα2 R149, IFNβ R1520 in (b)). Calculations and images were obtained with the


with ifnar2-EC, the association rate constants werevery similarly low for all IFNs, independent of therather different electrostatic potentials of the bind-ing sites (Figure 8(b)). Thus, the association of IFNsto ifnar1 seems not to be influenced by electrostaticsteering.Comparing relative affinities towards the recep-

tor subunits with the relative and differentialcellular activities, we discovered two strikingcorrelations: (i) relative ISGF3 activity and antiviralactivity were in good agreement with the relativeligand binding affinities towards ifnar2 and (ii) thedifferential antiproliferative potency correlatedwith the affinity towards ifnar1 relative to theaffinity to ifnar2. Based on these correlations, wesuccessfully engineered an IFNα2 mutant withextraordinary high differential antiproliferativepotency. Thus, we could clearly demonstrate thatthe affinities to the receptor subunits governdifferential signal activation, opening novel ther-apeutic prospects in engineering IFNs with specificactivity patterns. While our results cannot excludethat also the interaction kinetics is important fordifferential receptor activation, the equilibriumconstants proved sufficient for our correlations.With our current understanding of the receptorassembling mechanism, we can analyze the con-sequences of the differential affinities. We haverecently demonstrated that two pathways have tobe considered for ternary complex formation anddissociation (Figure 9(a)).37 The formation of binaryand ternary complexes depends not only on theequilibrium constants in solution (K1 and K4), butalso on the lateral, two-dimensional equilibriumconstants on the membrane (K2 and K3). Based onthe data determined for the individual interactionswith the receptor subunits and for ternary complexformation, we have used this model for simulatingligand binding and ternary complex formation fordifferent IFNs (Figure 9(b) and (c)) and for theIFNα2 mutants described above (Figure 9(d) and(e)) assuming two different effective cellular recep-tor surface concentrations. The receptor surfaceconcentrations were chosen according to experi-mental ligand binding affinities for the cell surfacereceptor. A characteristic feature of ligand bindingto two independent receptor subunits is thepossibility that a ligand molecule binds to each ofthe receptor subunits thereby obstructing theformation of a ternary complex. For this reason,ligand binding does not saturate at ratio of 1relative to the receptor concentration, and bell-shaped curves are observed for the ternary com-plex. Furthermore, these simulations reveal someimportant principles of how receptor surfaceconcentration affects ligand binding and ternarycomplex formation: (i) the affinity of the ligand forthe cell surface receptor decreases with the receptorsurface concentration; (ii) the contribution of ifnar1binding to the total affinity decreases with thereceptor surface concentration; (iii) the relativeamount of ligand on the surface involved in theternary complex decreases with decreasing receptor

surface concentrations. This effect strongly dependson the affinity to ifnar1, indicating that theconcentration, not the amount limits ternary com-plex formation. These results are in line with recentstudies showing that the activity pattern of IFNβcould be mimicked by engineering an IFNα2mutant with similar affinity towards ifnar1.29 Forthe IL4-receptor it was shown that indeed thereceptor concentration on the plasma membrane isa limiting factor for ternary complex formation.38

Thus, the surface concentration of the receptorsubunits plays a complex role for both ligandbinding and ternary complex formation. Strikingly,the cell surface concentrations of ifnar1 and ifnar2were shown to be differentially down-regulated bystimulation with different IFNs, depending on theiraffinity towards ifnar1.29 Receptor down-regulationis probably responsible for cell desensitization,which explains the high IFN concentrations requiredfor maintaining a long-term response such asantiproliferative activity. This is in line with thefact that cell lines with higher ifnar1 expressionlevels (such as DAUDI) are much more sensitive toIFNs in terms of antiproliferative activity. Wesuggest that differential desensitization by IFNswith different affinities to the receptor subunits maycontribute to differential signaling: for activation ofshort-term responses such as ISGF3 formation andantiviral activity, ligand binding to the surface is thelimiting parameter. For long-term responses such asantiproliferative response, however, differentialdown-regulation of the receptor expression on thecell surface may differentially affect the responsive-ness of the cells. Strikingly, IFNs with rathersymmetric affinities to the receptor subunits exhib-ited least differences between short and long-termactivity, suggesting that cellular responsiveness isleast affected by these ligands.

Methods

Protein biochemistry

IFNβ (formulated Rebif 22 μg and 44 μg) was obtainedfrom Serono GmbH, Unterschleißheim/Germany. IFNα2and ifnar2-EC with a C-terminal decahistidine-tag wereexpressed in Escherichia coli, refolded from inclusionbodies and purified by anion exchange and size-exclusionchromatography as described.36 Ifnar1-EC with a C-terminal decahistidine-tag was expressed in Sf9 insectcells and purified from the supernatant by immobilizedmetal ion affinity and size-exclusion chromatography asdescribed.26 IFNα1, IFNα8 and IFNα21 were amplified byPCR from human genomic DNA and cloned into the sameexpression plasmid pT7T3-18U containing a nine aminoacid residue cistron in front of the start codon in order toincrease expression efficiency.36,39 The proteins wereexpressed, refolded and purified as described forIFNα2.36 IFNτ2 was expressed in yeast and purified tohomogeneity as described.40,41 Site-specific mutationswere generated by ligase chain reaction. For site-specificfluorescence labelling, an additional cysteine was incor-porated by mutagenesis into the proteins at a position,

Figure 9. Ligand-induced ternary complex formation. (a) The pathways involved in ternary complex formation anddissociation. The concentrations of binary (B1, B2) and ternary (T) complexes depend on the ligand concentration, thesurface concentration of the receptor subunits, and the equilibrium constants K1–K4. At high ligand concentration, each ofthe receptor subunits will bind a separate ligand molecule (right).45 (b)–(e) Simulated concentration-dependentequilibrium ligand binding (dotted lines) and ternary complex formation (continuous lines) curves for wild-type IFNs ((b)and (c)) and mutants ((d) and (e)) used in this study. Stoichiometric amounts of ifnar1 and ifnar2 were assumed at surfaceconcentrations of 10−16 mol/mm2 ((b) and (d)) and 10−17 mol/mm2 ((c) and (e)). The amount of receptor-bound ligand(plotted as ordinate) was normalized to the concentration of the receptor subunits.


which was previously shown to be not important forligand binding (mutants IFNα2 S136C, IFNα8 S137C andIFNα21 S137C). The proteins were expressed and purifiedas the wild-type. After size-exclusion chromatography, theproteins were reacted overnight at 4 °C with a three tofivefold amount of Oregon Green 488 maleimide (Mole-cular Probes, Eugene). Excess dye and multiple-labelledIFN were removed by another step of anion exchangechromatography followed by reverse phase chromatogra-phy with a Sephasil C8 column (Amersham Biosciences).IFNω was expressed in Sf9 insect cells using baculovirusinfection. The vector pACgp67B harboring the gene ofmature IFNω cloned via the BamHI and EcoRI restrictionsites was co-transfected with linearized baculoviral DNA(BD Biosciences) and the baculovirus was producedaccording to the manufacturers instructions. For protein

expression, fresh Sf9 cell cultures (200 ml) were infectedand the supernatant was harvested three to four days afterinfection. The supernatant was adjusted to 20 mM Tris(pH 8.0), and was thoroughly dialyzed against 20 mM Tris(pH 8.0). After centrifugation, the supernatant wasapplied to a 1 ml cation exchange column (HiTrap SPHP, Amersham Biosciences). After washing with 20 mMTris (pH 8.0), the proteins were eluted with a gradientfrom 0–500 mM NaCl in 20 mM Tris (pH 8.0) (TBS). Thefractions containing IFNω were further purified by size-exclusion chromatography in TBS (Superdex 75–16/60,Amersham Biosciences). For site-specific fluorescencelabelling, the mutant IFNω S137C was generated by site-directed mutagenesis. After expression and purification asthe wild-type protein, the mutant was reduced with DTTprior to reaction with Oregon Green 488 maleimide. The


labelled protein was separated from the free dye by cationexchange chromatography.

Binding assays by solid phase detection

Label-free binding assays by reflectometric interfer-ence spectroscopy (RIfS) were carried out as des-cribed26,42 using a home-built set-up.43 Simultaneoustotal internal reflection fluorescence spectroscopy(TIRFS) and reflectance interference (RIf) detection wascarried out as described recently.27 All measurementswere carried out in 20 mM Hepes (pH 7.5), 150 mMNaCl (HBS). For monitoring the interaction withdifferent type I IFN, ifnar1-EC or ifnar2-EC wereimmobilized via their C-terminal decahistidine-tagsonto PEG-modified surfaces using multivalent chelatorsfor stable immobilization as described.26,33 Excesscoordination sites were blocked by injecting 1 μMdecahistidine-tagged maltose binding protein (MBP-H10) to avoid non-specific binding.33 Rate constantswere determined by fitting mono-exponential functionsas described. Equilibrium dissociation constants weredetermined from concentration-dependent equilibriumresponse by fitting a Langmuir isotherm. For competi-tion experiments, fluorescence-labelled and unlabelledIFNs were mixed in different ratios and injected. Ternarycomplex formation was measured with ifnar2-EC andifnar1-EC tethered onto solid-supported lipid bilayers asdescribed.26,27

Cellular activity assays

All activity assays were carried out with 2fTGH-cellsand its derivative HL116, which are stably transfectedwith an ISGF3-responsive luciferase reporter gene.2fTGH-cells were cultured in DMEM with 10% (v/v)FCS/ Hygromycin. HL116-cells were cultured in DMEMwith 10% FCS/HAT. IFN activity with respect to ISGF3activation in HL116 cells was determined as described.16

Briefly, cells were incubated with serial dilutions of IFNfor 6 h, and luciferase-activity was determined by addingluciferin after lysis of the cells using a luminescenceimager. EC50 values were determined by fitting asigmoidal curve to the dose-response curves. The anti-proliferative activity was measured by using 2fTGH-cellsand detected by serial dilution of different IFN as des-cribed before.29 Briefly, cells were incubated with serialdilutions of IFN for seven days. The amounts of viablecells were determined by staining with crystal violet anddetermining the absorption in an ELISA plate reader at620 nm. EC50 values were determined by fitting asigmoidal curve to the dose-response curves. As a control,antiproliferative activity assays were also carried out inthe same way with HeLa and HL116 cells. The antiviralactivity of different IFNs was quantified by using VSV.The inhibition of the cytopathic effect of VSV on 2fTHG-cells was assessed with serial dilution of different IFNs,and the concentration needed for 50% protection of thecells was determined.19 For the comparison of cellularactivities with binding affinities, the ISGF3, antiviral andantiproliferative potencies (pISGF3, pAV and pAP, respec-tively) were calculated as the reciprocal EC50 values andnormalized to the respective potency IFNα2 (relativepotencies prel

ISGF3, prelAV and prel

AP):

pISGF3 ¼ 1ECISGF3

50and pISGF3

rel ¼ pISGF3

pISGF3ðIFNa2Þ

The differential antiproliferative potency (pdiffAP ) was

determined as the relative antiproliferative potency prelAP

divided by the relative ISGF3 potency prelISGF3:

pAPdiff ¼

pAPrel

pISGF3rel

The relative binding affinities to the receptor subunitsifnar1 and ifnar2 (Krel

ifnar1 and Krelifnar2, respectively) were

calculated as the reciprocal equilibrium dissociationconstants KD and normalized to the respective reciprocalKD of IFNα2:

Kifnar1rel ¼ ðKifnar1

D Þ�1

Kifnar1D ðIFNa2Þ� ��1 and

Kifnar2rel ¼ ðKifnar2

D Þ�1

Kifnar2D ðIFNa2Þ� ��1

The affinity towards ifnar1 relative to the affinity towardsifnar2 (Krel

R1/R2) was determined as:

KR1=R2rel ¼ ðKifnar1

D =Kifnar2D Þ�1=2

Kifnar1D ðIFNa2Þ=Kifnar2

D ðIFNa2Þ� ��1=2

Simulation

Amount of receptor cross-linking and total ligandbinding was numerically simulated using the followingset of arithmetic equations:

K1 ¼ ½R2�½L�½B2� ; K2 ¼ ½B2�½R1�

½T� ; K4 ¼ ½R1�½L�½B1� ;

½R2�0 ¼ ½R2� þ ½B2� þ ½T�

½R1�0 ¼ ½R1� þ ½B1� þ ½T�½L�bound ¼ ½T� þ ½B2� þ ½B1�

where [L] is the ligand concentration in solution (in M−1);[R2] [R1] are the surface concentrations of the unoccupiedreceptor subunits ifnar2 and ifnar1, respectively; [B2], [B1]and [T] are the concentrations of the binary complexeswith the ligand and ternary complex, respectively aspresented in Figure 9(a) (measured in mol/mm2); [R2]0and [R1]0 are the total surface concentrations of receptorsubunits on the membrane; [L]bound is the total amount ofmembrane-bound ligand (measured in mol/mm2), whichis present in the forms of binary and ternary complexes.For the simulations, stoichiometric surface concentrationsof ifnar1 and ifnar2 were assumed, i.e. [R2]0=[R1]0. Theequilibrium constants K1 and K4 were taken from Table 1,and the two-dimensional equilibrium constant K2 wasestimated from K4 as described.27,38 For comparison ofdifferent receptor surface concentrations, [L]bound and [T]were normalized to the surface concentration of thereceptor subunits and plotted as a function of the ligandvolume concentration ([L]).

Acknowledgements

IFNβ (Rebif) was generously provided by DrGarth Virgin, Serono GmbH, Unterschleißheim. The


hospitality and the support from the laboratory ofRobert Tampé are gratefully acknowledged, inparticular excellent technical assistance by EckhardLinker with cell culture. This project was supportedby grants from the Deutsche Forschungsge-meinschaft (Emmy-Noether Program PI 405/1 andthe SFB 628) to J.P., and by the Human FrontierScience Program (RGP60/2002) to J.P. and G.S.

References

1. Belardelli, F. & Ferrantini, M. (2002). Cytokines as alink between innate and adaptive antitumor immu-nity. Trends Immunol. 23, 201–208.

2. Mogensen, K. E., Lewerenz, M., Reboul, J., Lutfalla, G.& Uze, G. (1999). The type I interferon receptor:structure, function, and evolution of a family business.J. Interferon Cytokine Res. 19, 1069–1098.

3. Pestka, S., Krause, C. D. & Walter, M. R. (2004).Interferons, interferon-like cytokines, and their recep-tors. Immunol. Rev. 202, 8–32.

4. Croze, E., Russell-Harde, D., Wagner, T. C., Pu, H.,Pfeffer, L. M. & Perez, H. D. (1996). The human type Iinterferon receptor. Identification of the interferonbeta-specific receptor-associated phosphoprotein.J. Biol. Chem. 271, 33165–33168.

5. Domanski, P., Nadeau, O. W., Platanias, L. C., Fish, E.,Kellum, M., Pitha, P. & Colamonici, O. R. (1998).Differential use of the betaL subunit of the type Iinterferon (IFN) receptor determines signaling speci-ficity for IFNalpha2 and IFNbeta. J. Biol. Chem. 273,3144–3147.

6. Rani, M. R., Foster, G. R., Leung, S., Leaman, D., Stark,G. R. & Ransohoff, R. M. (1996). Characterization ofbeta-R1, a gene that is selectively induced by inter-feron beta (IFN-beta) compared with IFN-alpha.J. Biol. Chem. 271, 22878–22884.

7. Abramovich, C., Shulman, L. M., Ratovitski, E.,Harroch, S., Tovey, M., Eid, P. & Revel, M. (1994).Differential tyrosine phosphorylation of the IFNARchain of the type I interferon receptor and of anassociated surface protein in response to IFN-alphaand IFN-beta. EMBO J. 13, 5871–5877.

8. Platanias, L. C., Uddin, S., Domanski, P. &Colamonici, O. R. (1996). Differences in interferonalpha and beta signaling. Interferon beta selectivelyinduces the interaction of the alpha and betaLsubunits of the type I interferon receptor. J. Biol.Chem. 271, 23630–23633.

9. Coelho, L. F., de Freitas Almeida, G. M., Mennechet,F. J., Blangy, A. & Uze, G. (2005). Interferon-{alpha}and -{beta} differentially regulate osteoclastogenesis:role of differential induction of chemokine CXCL11expression. Proc. Natl Acad. Sci. USA, 102,11917–11922.

10. Deonarain, R., Chan, D. C., Platanias, L. C. & Fish,E. N. (2002). Interferon-alpha/beta-receptor interac-tions: a complex story unfolding. Curr. Pharm. Des. 8,2131–2137.

11. Subramaniam, P. S., Larkin, J., 3rd, Mujtaba, M. G.,Walter, M. R. & Johnson, H. M. (2000). The COOH-terminal nuclear localization sequence of interferongamma regulates STAT1 alpha nuclear translocation atan intracellular site. J. Cell Sci. 113, 2771–2781.

12. da Silva, A. J., Brickelmaier, M., Majeau, G. R.,Lukashin, A. V., Peyman, J., Whitty, A. & Hochman,P. S. (2002). Comparison of gene expression patterns

induced by treatment of human umbilical veinendothelial cells with IFN-alpha 2b vs. IFN-beta 1a:understanding the functional relationship betweendistinct type I interferons that act through a commonreceptor. J. Interferon Cytokine Res. 22, 173–188.

13. Der, S. D., Zhou, A., Williams, B. R. & Silverman, R. H.(1998). Identification of genes differentially regulatedby interferon alpha, beta, or gamma using oligo-nucleotide arrays. Proc. Natl Acad. Sci. USA, 95,15623–15628.

14. de Veer, M. J., Holko, M., Frevel, M., Walker, E., Der,S., Paranjape, J. M. et al. (2001). Functional classifica-tion of interferon-stimulated genes identified usingmicroarrays. J. Leukoc. Biol. 69, 912–920.

15. Pearce, K. H., Jr, Cunningham, B. C., Fuh, G., Teeri, T.&Wells, J. A. (1999). Growth hormone binding affinityfor its receptor surpasses the requirements for cellularactivity. Biochemistry, 38, 81–89.

16. Lewerenz, M., Mogensen, K. E. & Uze, G. (1998).Shared receptor components but distinct complexesfor alpha and beta interferons. J. Mol. Biol. 282,585–599.

17. Piehler, J. & Schreiber, G. (1999). Mutational andstructural analysis of the binding interface betweentype I interferons and their receptor ifnar2. J. Mol. Biol.294, 223–237.

18. Runkel, L., Pfeffer, L., Lewerenz, M., Monneron, D.,Yang, C. H., Murti, A. et al. (1998). Differences inactivity between alpha and beta type I interferonsexplored by mutational analysis. J. Biol. Chem. 273,8003–8008.

19. Piehler, J., Roisman, L. C. & Schreiber, G. (2000). Newstructural and functional aspects of the type Iinterferon-receptor interaction revealed by compre-hensive mutational analysis of the binding interface.J. Biol. Chem. 275, 40425–40433.

20. Runkel, L., deDios, C., Karpusas, M., Betzenhauser,M., Muldowney, C., Zafari, M. et al. (2000). Systematicmutational mapping of sites on human interferon-beta-1a that are important for receptor binding andfunctional activity. Biochemistry, 39, 2538–2551.

21. Lamken, P., Lata, S., Gavutis, M. & Piehler, J. (2004).Ligand-induced assembling of the type I interferonreceptor on supported lipid bilayers. J. Mol. Biol. 341,303–318.

22. Roisman, L. C., Piehler, J., Trosset, J. Y., Scheraga,H. A. & Schreiber, G. (2001). Structure of theinterferon-receptor complex determined by distanceconstraints from double-mutant cycles and flexibledocking. Proc. Natl Acad. Sci. USA, 98, 13231–13236.

23. Cutrone, E. C. & Langer, J. A. (2001). Identification ofcritical residues in bovine IFNAR-1 responsible forinterferon binding. J. Biol. Chem. 276, 17140–17148.

24. Cajean-Feroldi, C., Nosal, F., Nardeux, P. C., Gallet,X., Guymarho, J., Baychelier, F. et al. (2004).Identification of residues of the IFNAR1 chain ofthe type I human interferon receptor critical forligand binding and biological activity. Biochemistry,43, 12498–12512.

25. Lamken, P., Gavutis, M., Peters, I., Van der Heyden, J.,Uze, G. & Piehler, J. (2005). Functional cartography ofthe extracellular domain of the type I interferonreceptor subunit ifnar1. J. Mol. Biol. 350, 476–488.

26. Lamken, L., Lata, S., Gavutis, M. & Piehler, J. (2004).Ligand-induced assembling of the type I interferonreceptor on supported lipid bilayers. J. Mol. Biol. 341,303–318.

27. Gavutis, M., Lata, S., Lamken, P., Müller, P. & Piehler,J. (2005). Lateral ligand-receptor interactions on


membranes probed by simultaneous fluorescence-interference detection. Biophys. J. 88, 4289–4302.

28. Hu, R., Bekisz, J., Schmeisser, H., McPhie, P. & Zoon,K. (2001). Human IFN-alpha protein engineering: theamino acid residues at positions 86 and 90 areimportant for antiproliferative activity. J. Immunol.167, 1482–1489.

29. Jaitin, D. A., Roisman, L. C., Jaks, E., Gavutis, M.,Piehler, J., Van der Heyden, J. et al. (2006). Inquiringinto the differential action of interferons (IFNs): anIFN-{alpha}2 mutant with enhanced affinity toIFNAR1 is functionally similar to IFN-{beta}. Mol.Cell. Biol. 26, 1888–1897.

30. Assal-Meliani, A., Charpigny, G., Reinaud, P., Martal,J. & Chaouat, G. (1993). Recombinant ovine tropho-blastin (roTP) inhibits ovine, murine and humanlymphocyte proliferation. J. Reprod. Immunol. 25,149–165.

31. Subramaniam, P. S., Khan, S. A., Pontzer, C. H. &Johnson, H. M. (1995). Differential recognition of thetype I interferon receptor by interferons tau and alphais responsible for their disparate cytotoxicities. Proc.Natl Acad. Sci. USA, 92, 12270–12274.

32. Martal, J. L., Chene, N. M., Huynh, L. P., L'Haridon,R. M., Reinaud, P. B., Guillomot, M. W. et al. (1998).IFN-tau: a novel subtype I IFN1. Structural char-acteristics, non-ubiquitous expression, structure-func-tion relationships, a pregnancy hormonal embryonicsignal and cross-species therapeutic potentialities.Biochimie, 80, 755–777.

33. Lata, S. & Piehler, J. (2005). Stable and functionalimmobilization of histidine-tagged proteins via multi-valent chelator head-groups on a molecular poly(ethylene glycol) brush. Anal. Chem. 77, 1096–1105.

34. Lu, J., Chuntharapai, A., Beck, J., Bass, S., Ow, A., DeVos, A. M. et al. (1998). Structure-function study of theextracellular domain of the human IFN-alpha receptor(hIFNAR1) using blocking monoclonal antibodies: therole of domains 1 and 2. J. Immunol. 160, 1782–1788.

35. Roisman, L. C., Jaitin, D., Baker, D. P. & Schreiber, G.(2005). Mutational analysis of the IFNAR1 binding site

on IFNalpha2 reveals the architecture of a weakligand-receptor binding-site. J. Mol. Biol. 353, 271–281.

36. Piehler, J. & Schreiber, G. (1999). Biophysical analysisof the interaction of human ifnar2 expressed in E.coliwith IFN alpha 2. J. Mol. Biol. 289, 57–67.

37. Gavutis, M., Jaks, E., Lamken, P. & Piehler, J. (2006).Determination of the 2-dimensional interaction rateconstants of a cytokine receptor complex. Biophys. J.90, 3345–5335.

38. Whitty, A., Raskin, N., Olson, D. L., Borysenko, C. W.,Ambrose, C. M., Benjamin, C. D. & Burkly, L. C.(1998). Interaction affinity between cytokine receptorcomponents on the cell surface. Proc. Natl Acad. Sci.USA, 95, 13165–13170.

39. Schoner, B. E., Belagaje, R. M. & Schoner, R. G. (1990).Enhanced translational efficiency with two-cistronexpression system. Methods Enzymol. 185, 94–103.

40. Martal, J., Degryse, E., Charpigny, G., Assal, N.,Reinaud, P., Charlier, M. et al. (1990). Evidence forextended maintenance of the corpus luteum byuterine infusion of a recombinant trophoblast alpha-interferon (trophoblastin) in sheep. J. Endocrinol. 127,R5–R8.

41. Degryse, E., Dietrich, M., Nguyen, M., Achstetter, T.,Charlier, M., Charpigny, G. et al. (1992). Addition of adipeptide spacer significantly improves secretion ofovine trophoblast interferon in yeast. Gene, 118,47–53.

42. Piehler, J. & Schreiber, G. (2001). Fast transientcytokine-receptor interactions monitored in real timeby reflectometric interference spectroscopy. Anal.Biochem. 289, 173–186.

43. Schmitt, H. M., Brecht, A., Piehler, J. & Gauglitz, G.(1997). An integrated system for optical biomolecularinteraction analysis. Biosens. Bioelect. 12, 809–816.

44. Guex, N. & Peitsch, M. C. (1997). SWISS-MODELand the Swiss-PdbViewer: an environment forcomparative protein modeling. Electrophoresis, 18,2714–2723.

45. Whitty, A. & Borysenko, C. W. (1999). Small moleculecytokine mimetics. Chem. Biol. 6, R107–R118.

Edited by I. Wilson

(Received 27 August 2006; received in revised form 15 November 2006; accepted 15 November 2006)Available online 18 November 2006

Differential Receptor Subunit Affinities of Type I …...Type I interferons (IFNs) elicit antiviral,...

Documents

Transcript of Differential Receptor Subunit Affinities of Type I …...Type I interferons (IFNs) elicit antiviral,...