Ligand Binding Induces a Conformational Change in ifnar1 ... JMB2008.… · Ligand Binding Induces a Conformational Change in ifnar1 that Is Propagated to Its Membrane-Proximal Domain

doi:10.1016/j.jmb.2008.01.017 J. Mol. Biol. (2008) 377, 725–739

Available online at www.sciencedirect.com

Ligand Binding Induces a ConformationalChange in ifnar1 that Is Propagated to ItsMembrane-Proximal Domain

Jennifer Julia Strunk1, Ingo Gregor2, Yvonne Becker1, Zongli Li3,Martynas Gavutis1, Eva Jaks1, Peter Lamken1, Thomas Walz3,Jörg Enderlein4 and Jacob Piehler1⁎

1Institute of Biochemistry,Johann Wolfgang Goethe-University, Biocenter N210,Max-von-Laue-Straße 9, 60438Frankfurt am Main, Germany2Institute of Neurosciences andBiophysics 1, Research CenterJülich, 52425 Jülich, Germany3Department of Cell Biology,Harvard Medical School,Boston, MA 02115, USA4Institute of Physical andTheoretical Chemistry,Eberhard-Karls-University,Tübingen, 72076 Tübingen,Germany

Received 4 October 2007;received in revised form20 December 2007;accepted 7 January 2008Available online16 January 2008

*Corresponding author. E-mail [email protected] used: IFN, interfero

resonance energy transfer; ifnar1-ECifnar1; H10, decahistidine; OG488, Oelectron microscopy; RIf, reflectancetotal internal reflection fluorescenceHepes-buffered saline; EDTA, ethyleacid; NTA, nitrilotriacetic acid.

0022-2836/$ - see front matter © 2008 E

The type I interferon (IFN) receptor plays a key role in innate immunityagainst viral and bacterial infections. Here, we show by intramolecularFörster resonance energy transfer spectroscopy that ligand binding inducessubstantial conformational changes in the ectodomain of ifnar1 (ifnar1-EC).Binding of IFNα2 and IFNβ induce very similar conformations of ifnar1,which were confirmed by single-particle electron microscopy analysis of theternary complexes formed by IFNα2 or IFNβwith the two receptor subunitsifnar1-EC and ifnar2-EC. Photo-induced electron-transfer-based fluores-cence quenching and single-molecule fluorescence lifetime measurementsrevealed that the ligand-induced conformational change in the membrane-distal domains of ifnar1-EC is propagated to its membrane-proximaldomain, which is not involved in ligand recognition but is essential forsignal activation. Temperature-dependent ligand binding studies as well asstopped-flow fluorescence experiments corroborated a multistep conforma-tional change in ifnar1 upon ligand binding. Our results thus suggest thatthe relatively intricate architecture of the type I IFN receptor complex isdesigned to propagate the ligand binding event to and possibly even acrossthe membrane by conformational changes.

© 2008 Elsevier Ltd. All rights reserved.

Keywords: type I interferon receptor; fluorescence spectroscopy; cytokinereceptor; protein–protein interaction; conformational dynamics
Edited by I. B. Holland
Introduction

Type I interferons (IFNs) are cytokines with keyfunctions in the innate immune response against

ess:

n; FRET, Förster, ectodomain ofregon Green 488; EM,interference; TIRFS,spectroscopy; HBS,nediaminetetraacetic

lsevier Ltd. All rights reserve

intracellular infections and malignancies.1–3 Con-sequently, IFNs have substantial medical potentialand are already used in the treatment of hepatitis Cand multiple sclerosis as well as several kinds ofcancer.4 All type I IFNs bind to the same cell surfacereceptor, which is composed of the two subunitsifnar1 and ifnar2.5 Strikingly, different members ofthe IFN family instigate different cellular responseswhile using the same cell-surface receptor.6–11 Abetter structural and functional understanding oftype I IFN receptor assembly could therefore pavethe road for the rational engineering of IFNs withactivity patterns designed for medical applications.Over the past years, mutational and NMR studies

have systematically unraveled the recognition of IFNs

d.

mailto:[email protected]

726 Conformational Changes in ifnar1

by their two receptor subunits.12–20 IFNs bind withnanomolar affinity to an epitope on ifnar2, whichshows the architecture of a typical cytokine-bindingmodule composed of two Ig-like domains.15,19,21In contrast, the three N-terminal domains of ifnar1are required for binding of IFNs,22,23 yet ifnar1 bindsIFN ligands with affinities that are 2–3 orders ofmagnitude lower than that of ifnar2.24 Mutationalstudies have corroborated the involvement of thethree N-terminal Ig-like domains of ifnar1 in theinteraction with the ligand,25,26 which was alsoconfirmed by a recent three-dimensional recons-truction of the ternary complex by a single-particleelectron microscopic analysis.27 The membrane-proximal Ig-like domain is not required for ligandbinding, but has very specific properties requiredfor the assembly of the ternary complex on theplasma membrane.23 In the three-dimensional re-construction of the ternary complex, the membrane-proximal domains of ifnar1 and ifnar2 are in closeproximity.27 While the role of the unusual architec-ture of the interaction between ifnar1 and IFNsremains enigmatic, a possible explanation would bethe differential use of ifnar1 by different IFNs, whichmay also explain the observed differences in signalactivation. Interactions with ligands mediated bymultiple receptor domains and the resulting largebinding interface points, however, also towards aconformational change in ifnar1. Ligand-inducedconformational changes have already been suggestedto play a critical role in cytokine receptor assemblyand signaling,28–37 but the underlying mechanismshave so far not been explored experimentally. NMRexperiments have revealed only minor conforma-tional changes in ifnar2-EC and IFNα2 upon complexformation. Binding studies have established that thetwo receptor subunits interact with the ligandindependently of each other.17,20,38 The distributionof the ligand-binding site of ifnar1 over three Ig-likedomains23 suggests that ligand binding may cause arearrangement in the overall structure of ifnar1. Thekey role of ifnar1 recognition for mediating differ-ential responses has been demonstrated.24,39,40 Thus,differences in the recruitment of IFN ligand and theresulting conformational changes in ifnar1 mayaccount for the different cellular responses to thebinding of different IFNs.Here, we used several fluorescence techniques

and single-particle electron microscopy (EM) to ex-plore the conformation of the ectodomain of ifnar1(ifnar1-EC) in complex with ifnar2-EC and two dif-ferent IFNs. We observed a substantial spatial re-arrangement of the Ig-like domains of ifnar1-ECupon ligand binding. Strikingly, this conformationalchange is propagated from the membrane-distaldomains to the membrane-proximal domain ofifnar1-EC, which is not involved in ligand binding.The conformational change in ifnar1 resulting frombinding of IFNα2 and IFNβ was indistinguishable,suggesting that its propagation to the membrane-proximal domain is conserved and may play animportant role for receptor assembly and transmem-brane signaling.

Results

Ligand-induced conformational changes inifnar1-H10

We hypothesized that the interaction of multipleIg-like domains of ifnar1-EC with an IFN ligandcould induce a change in the spatial organizationof the involved receptor domains. In order to test thisidea, we first probed the conformation of ifnar1-ECby Förster resonance energy transfer (FRET) experi-ments. For this assay, the donor fluorophore wascovalently coupled to cysteine residues, whichwere incorporated into ifnar1-EC by site-directedmutagenesis. The quencher fluorophore was intro-duced in situ by noncovalent interactions with theC-terminal decahistidine (H10) tag using the multi-valent chelator tris-nitrilotriacetic acid (NTA)41,42 asschematically depicted in Fig. 1a and b. Detailedcharacterization and calibration of this technique,which will be published elsewhere, confirmed theFRET-based quenching of the donor fluorophore.Ifnar1-EC mutants with individual cysteinesinserted into different Ig-like domains (cf. Fig. 1c)were site-specifically labeled with Oregon Green 488(OG488) maleimide and purified to homogeneity.The Cy5 analogue fluorophore FEW646 was used asFRET acceptor and conjugated to the multivalentchelator tris-NTA (FEW646tris-NTA, Fig. 1a). It hasbeen shown that such conjugates bind with highaffinity and selectivity to His-tags introduced intoproteins,41 thus yielding quantitative labeling withthe acceptor dye.Upon addition of FEW646tris-NTA to the OG488-

labeled ifnar1-H10 cysteine mutants (OG488ifnar1-H10), a strong decrease in fluorescence intensity wasobserved, as shown for OG488ifnar1-H10N23C in Fig.1d and OG488ifnar1-H10 N349C in Fig. 1e. The extentof fluorescence quenching depended on the positionwhere OG488 was inserted in ifnar1-H10 andcorrelated with the distance of the labeled cysteineresidue from the C-terminal His tag. For the S147C,N286C and N349C mutants, the relative quenchingamplitudes were qualitatively in agreement with thedistances of the mutated residues from the C ter-minus as estimated from the structural model of theternary complex27 (Fig. 1c). For the N23C mutant,however, the quenching amplitude was similar tothose of the S147C and N286C mutants, which is notconsistent with the larger distance of this residuefrom the C terminus compared to the other tworesidues (Fig. 1c). These results suggest that ifnar1-EC in the absence of a bound ligand adopts a bentconformation. The similar distances of the threeresidues N23, S147 and N286 from the C terminus inthis conformation would explain the similar extentsof fluorescence quenching observed in the FRETexperiments. If free ifnar1 is bent, while extended inthe ternary ifnar1-EC–ifnar2-EC–IFNα2 complex,ligand binding should cause a measurable changein fluorescence quenching with the N23C mutant.Indeed, addition of the ligand IFNα2-HEQ (which

Fig. 1. FRET analysis of the conformation of ifnar1-H10 in solution. (a) Chemical structure of FEW646tris-NTA andschematic drawing of its multivalent interaction with a histidine tag. (b) Schematic representation of the fluorescenceassay: in situ labeling of OG488ifnar1-H10 N23C with FEW646tris-NTA for analysis of the conformation of free receptor bydonor quenching followed by ligand binding and monitoring dequenching of the donor as a result of changes in distance.(c) Model of the ternary complex formed by IFNα2 with the ectodomains of ifnar1 and ifnar2 based on a single-particleEM reconstruction.27 (d) Fluorescence spectrum of OG488ifnar1-H10 N23C before (black) and after (red) addition ofFEW646tris-NTA, and upon binding of IFNα2-HEQ (blue). (e) Same experiment as shown in (a) carried out withOG488ifnar1-H10 N349C. Competition with tag-less ifnar1 (ifnar1-tl) is shown in green. (f) Comparison of the relativefluorescence quenching for ifnar1-EC labeled with OG488 at different positions. (g) Comparison of the relativefluorescence quenching of OG488ifnar1-H10 N23C with AT565tris-NTA and fluorescence recovery after addition of IFNαand IFNβ. (h) Zoom into the membrane-proximal domain of ifnar1-EC (green) within the structural model of the ternarycomplex with IFNα2 (not visible) and ifnar2-EC (depicted in blue).27 N349 is depicted in orange, W347 in magenta.

727Conformational Changes in ifnar1

binds ifnar1-EC with∼100 nM affinity) at saturatingconcentrations resulted in a substantial recovery influorescence (Fig. 1d). A significant recovery influorescence was, however, not only observed forOG488ifnar1-H10 N286C, but, surprisingly, also forOG488ifnar1-H10 N349C (Fig. 1e and f), supporting asubstantial rearrangement of the molecule uponinteraction with the ligand. While the fluorescencerecovery pattern obtained for the mutants N23, S147and N286 (Fig. 1f) points towards a move of theN-terminal cytokine-binding module along thelinker between Ig-like domains 2 and 3, the effecton OG488ifnar1-H10 N349C was explained by theproximity to the Trp residue 347 (Fig. 1h) as de-monstrated by further experiments (see below). Thestrongest recovery was observed for the N23C mu-tant, and the resulting quenching amplitudes after

ligand bindingwere qualitatively consistent with thedistances of the residues in the model of the ternarycomplex (Fig. 1c). Taking into account the theoreticalFörster radius of ∼50 Å for this FRET pair, weestimate that ligand binding increases the distance ofN23 from the C terminus by ∼13 Å.The ligand-induced rearrangement of the Ig-like

domains was confirmed with a different acceptorfluorophore, AT565tris-NTA,41 for which both stron-ger quenching of the donor fluorophore as well ashigher recovery upon ligand binding was observed(Fig. 1g). These measurements yielded a similarincrease in the distance of ∼11 Å between N23 andthe C terminus upon ligand binding. Using thismoresensitive probe, we furthermore compared the fluo-rescence dequenching of OG488ifnar1-H10 N23Cupon binding of IFNα2-HEQ in complex with


ifnar2-EC, and of IFNβ in complex with ifnar2-EC.These experiments revealed that the ligand-inducedconformation of ifnar1 is (i) independent of ifnar2and (ii) independent of the type of ligand (Fig. 1g).

Corresponding architectures of the ternarycomplexes formed with IFNα2 and IFNβ

The results described above suggested that bind-ing of IFNα2 and IFNβ, which cause the mostdifferent cellular responses, induces very similarconformations of ifnar1-EC. We therefore comparedthe architectures of the ternary signaling complexesformed by IFNα2 and IFNβ by single-particle EManalysis. The stable complex of IFNβ with ifnar1-H10 and ifnar2-H10 was isolated by size-exclusionchromatography and imaged by negative stain EM.Individual complexes could be clearly identified(Fig. 2a). A total of 14,782 particles were selectedinteractively from the images and classified into 50classes (Supplementary Fig. 1). A typical classaverage is shown in Fig. 2b in comparison with acorresponding class average previously obtained forthe ternary complex formed with IFNα2-HEQ (Fig.2c).27 The arrangement of the Ig-like domains ofifnar1-EC in the ternary complex formed with IFNβappeared very similar to that seen in the complexformed with IFNα2-HEQ. The relative orientation ofthe receptor subunits is also indistinguishable in thetwo IFN complexes. Remarkably, a similar relativeshift between the membrane-proximal domains ofifnar1-EC and ifnar2-EC can be observed in bothternary complexes. The similar mode of binding ofIFNα2 and IFNβ to ifnar1 was confirmed by quanti-

tative ligand-binding studies with ifnar1 single-amino-acid mutants using reflectance interference(RIf) detection (Table 1). All residues involved in theinteractions with IFNα2-HEQ were also involved inthe interaction with IFNβ. These results corroboratethat IFNα2 and IFNβ form ternary complexes withvery similar architectures.

Conformational changes are propagated fromthe membrane-distal domains to themembrane-proximal domain of ifnar1-EC

A very surprising outcome of the fluorescencequenching assays shown in Fig. 1 was the ratherstrong dequenching observed for OG488ifnar1-H10N349C (Fig. 1e). In case of this mutant, OG488 isattached to the C-terminal Ig-like domain, whichcarries the His tag and thus the quencher dye.Furthermore, binding studies with various deletionand chimeric constructs23 as well as the structure ofthe ternary complex obtained by single-particleEM27 suggested that this domain is not involved ininteractions with ligands. Furthermore, the ligand-induced fluorescence dequenching of OG488ifnar1-H10 N349C was largely independent on the quen-cher attached to the His tag (Supplementary Fig. 2).To explore the conformation of the C-terminal Ig-likedomain in more detail, we attached the fluorophoreATTO 655 to ifnar1-H10 N349C (AT655ifnar1-H10N349C). ATTO 655 is an oxazine-based dye that isvery potently quenched by photo-induced electrontransfer to tryptophan.43 We speculated thatW347, atryptophan residue in the vicinity of N349, may beable to quench the covalently attached fluorophore.

Fig. 2. Conformation of ifnar1-EC in ternary IFN–receptor com-plexes as revealed by single-particleEM analysis. (a) Image of a nega-tively stained sample of ternarycomplexes formed by IFNβ withifnar2-EC and ifnar1-EC. Some ofthe particles are marked by circles.Scale bar is 50 nm. Selected classaverage of the ternary complexformed with IFNβ (b) in compar-ison with a corresponding classaverage of the ternary complexformed with IFNα2-HEQ (c).27

Table 1. Binding affinities of ifnar1-EC mutants forIFNα2-HEQ and IFNβ

Mutant

KD

IFNα2-HEQ IFNβ

Wild type 70±20 nM 50±20 nMY70A N3 μM N500 nMR76A Like wild type Like wild typeE111A 340±150 nM N2 μMK113A N3 μM N2 μMW129A N3 μM N2 μMF136A N3 μM N2 μMN155T 200±40 M 1±0.3 nML247A Like wild type Like wild typeW250A —a —a

K251A 130±20 nM Like wild-typeD298A Like wild type Like wild typeN368D Like wild type Like wild type

a No monomeric protein was obtained for this ifnar1 mutant.


Indeed, a more than twofold increase in fluorescenceof AT655ifnar1-H10 N349C was observed uponbinding of the ligand IFNα2-HEQ (Fig. 3a). Further-more, the peak position shifted slightly upon ligandbinding,which is typical for quenching of this dye bytryptophan.43 The fluorescence quenching of ATTO655was shown to be due to interactionwithW347, asno ligand-induced increase in fluorescence wasobserved for the W347F, N349C double mutant ofAT655ifnar1-H10 (Fig. 3b). Unchanged ligand-bindingactivity of ifnar1-H10 W347F, N349C was confirmedby solid-phase binding assays using RIf detection(data not shown). Moreover, ligand-induced de-quenching of AT655ifnar1-H10 N349Cwas monitoredin real time by simultaneous total internal reflectionfluorescence spectroscopy (TIRFS)–RIf detection(Fig. 3c and d). The fluorescence and mass-sensitivesignals were in excellent agreement (Fig. 3d).With this very sensitive probe for conformational

changes in hand, we analyzed the influence ofifnar2-EC on the conformation of the binary ifnar1-EC/IFNα2 complex and the potential differencesbetween IFNα2 and IFNβ (Fig. 3e and Supplemen-tary Fig. 2). The dequenching amplitudes showed nostatistically significant differences for IFNα2, IFNα2in complex with ifnar2-EC, and IFNβ in complexwith ifnar2-EC. Since the membrane-proximaldomain is not involved in ligand binding, theseresults suggest that the ligand-induced rearrange-ment of the three N-terminal Ig-like domains ofifnar1 is propagated to the membrane-proximal do-main. Furthermore, binding of IFNα2 and IFNβresults in a very similar conformational change ofthe membrane proximal domain of ifnar1-EC, con-firming that ifnar1 adopts very similar conforma-tions in complex with IFNα2 and IFNβ.

Accessibility of W347 is drastically reducedupon ligand binding

Although the temperature dependence of thefluorescence intensity of AT655ifnar1-H10 N349Cwas negligible (Fig. 3f), indicating that conforma-tional flexibility contributed little to fluorescence

dequenching, these measurements did not comple-tely rule out the possibility that loss of conforma-tional flexibility may play a role in dequenching.Therefore, we further explored the nature of theconformational change in the membrane-proximaldomain of ifnar1 by time-resolved fluorescencemeasurements. Ensemble fluorescence lifetime mea-surements of AT655ifnar1-H10 N349C in the absenceand presence of IFNα2-HEQ indicated only littlechange of the fluorescence lifetime of ATTO 655 (Fig.4a), confirming a fast transition of the fluorophorebetween a fully quenched and an unquenched stateas expected for quenching processes based onelectron transfer. The transition dynamics of thefluorophore were further characterized by fluores-cence correlation spectroscopy. For this purpose, thefluorescence of ∼1 nM AT655ifnar1-H10 N349C wasmonitored in the absence and presence of the ligandIFNα2-HEQ by time-correlated single-photon count-ing. The autocorrelation functions obtained for lagtime τ ranged from 25 ns to 1 s (Fig. 4b and c).Without ligand, two decay times could be clearlydistinguished in the autocorrelation function: one inthe millisecond range, which corresponds to thediffusion time, and one in the microsecond range,which is attributed to the dynamic quenching pro-cess, which requires the fluorophore to come intovery close proximity to the tryptophan residue.Strikingly, ligand binding almost completely elimi-nated the decay of the autocorrelation in the micro-second range (Fig. 4b). The same measurementswere carried out with the AT655ifnar1-H10 doublemutantW347F, N349C. The absence of the fast decayin the microsecond range in the absence of W347(Fig. 4c) confirmed that this tryptophan residuewas responsible for the dynamic quenching ofAT655ifnar1-H10 N349C. These results thus corrobo-rated that ligand binding causes a shift of themembrane-proximal domain of ifnar1. The fastautocorrelation time of ∼5 μs corresponds wellwith the autocorrelation time measured for a flexibletryptophan-containing peptide linked to an analo-gous dye,43 suggesting that the quenching dynamicsare determined by the conformational dynamics ofthe linker between the dye and the protein. The factthat the fast autocorrelation almost completelydisappeared upon ligand binding indicates that theaccessibility of W347 is substantially changed. Thesesingle-molecule measurements therefore support aligand-induced conformational change rather than aloss in flexibility.

Non-Arrhenius temperature-dependentligand-binding kinetics of ifnar1-EC

In order to explore the mechanistic implications ofthe ligand-induced conformational change, wemeasured the temperature dependency of the rateconstants of complex formation for ifnar1-H10 incomparison to ifnar2-H10. The proteins were cap-tured onto NTA-functionalized gold surfaces, andbinding of IFNα2 and IFNα2-HEQwasmonitored atvarious temperatures between 5 and 40 °C by surface

Fig. 3. Conformational changes in AT655ifnar1-H10 N349C detected by electron-transfer-based fluorescence quenching.(a) Fluorescence spectra of AT655ifnar1-H10 N349C before and after addition of IFNα2-HEQ. (b) Same experiment asshown in (a) with ATTO655ifnar1-H10W347F, N349C. (c) Binding of IFNα2-HEQ to immobilized ATTO655ifnar1-H10 N349Cas detected by simultaneous TIRFS (red curve) and RIf (black curve) measurements. (d) Comparison of the normalizedsignals during ligand binding detected by TIRFS (red curve) and by RIf (black curve). (e) Comparison of the relativechanges in fluorescence upon binding of IFNα2-HEQ, IFNα2-HEQ in complex with ifnar2-EC and IFNβ in complex withifnar2-EC to AT655ifnar1-H10 N349C. (f) Temperature dependence of the peak fluorescence intensity of ATTO655ifnar1-H10N349C.


plasmon resonance (full binding curves at selectedtemperatures are shown in Supplementary Fig. 3,and the rate constants are summarized in Table 2).For ifnar2, the rate constants increased substantiallywith higher temperatures, which is particularlyapparent in the dissociation curves (Fig. 5a). Theassociation and dissociation rate constants obtainedfrom these curves are fully consistent with the

Arrhenius law (Fig. 5c and d), yielding activationenergies of 60 kJ mol−1 for the association kineticsand 85 kJ mol−1 for the dissociation kinetics. Thesame results were obtained for IFNα2-HEQ bindingto ifnar2-H10 (data not shown). In contrast, adecrease in the dissociation rate constant wasobserved for ligand binding to immobilized ifnar1-H10 (Fig. 5b). The dissociation rate constant only

Fig. 4. Conformational changes of ifnar1-EC as mon-itored by time-resolved fluorescence. (a) Ensemble fluor-escence lifetime measurements of AT655ifnar1-H10 N349Cin the absence and presence of the ligand in comparison tofree ATTO 655 dye. (b) Autocorrelation curves forAT655ifnar1-H10 N349C in the absence and presence ofthe ligand. (c) Autocorrelation curves for AT655ifnar1-H10W347F, N349C in the absence and presence of the ligand.

Table 2. Rate and equilibrium constants obtained withsolid-phase binding assays at different temperatures

Ifnar2-H10/IFNα2Ifnar1-H10/IFNα2-HEQ

T(°C)ka

(M−1 s−1)akd

(s−1)bKD

(nM)cka

(M−1 s−1)kd

(s−1)KD(nM)

5 7.8×105 0.0023 3.0 5.6×104 0.13 230010 1.3×106 0.0042 3.2 7.7×104 0.092 120015 2.1×106 0.0072 3.4 1.5×105 0.067 43020 3.2×106 0.015 4.6 1.8×105 0.051 28025 5.1×106 0.024 4.7 3.1×105 0.047 15030 7.1×106 0.042 5.9 3.4×105 0.048 14035 1.0×107 0.082 8.2 3.5×105 0.060 17040 1.6×107 0.16 9.9 3.9×105 0.087 220

a Association rate constant (average standard deviation of∼15%).

b Dissociation rate constant (average standard deviation of∼10%).

c Equilibrium dissociation constant determined from the rateconstants (average standard deviation of ∼17%).


began to increase again at a temperature above 30 °C.In the Arrhenius plots, both association and dis-sociation kinetics behaved in a nonlinear fashion(Fig. 5c and d). The slope of theArrhenius plot for theassociation rate constants changed significantlybetween 20 and 25 °C, but the dissociation rateconstants showed a positive slope between 5 and

30 °C, yielding an apparent negative activationenergy in this temperature range. Very similar resultswere obtained with ifnar1-EC without the mem-brane-proximal Ig-like domain (H10-SD123). Inprotein folding, apparent negative activation ener-gies have been attributed to heat-capacity changes inthe activated complex or to temperature-inducedchanges of the ground state.44 For the interaction offolded proteins, however, apparent negative activa-tion energies have been explained by changes in therate-limiting step of multistep reactions.45 The tem-perature dependence of the kinetics of the ifnar1–IFNα2 complex formation indicates a transition ofthe rate-limiting step between 20 and 30 °C. To ourknowledge, negative activation energies have onlybeen reported for the association of macromolecularcomplexes that cause large conformational changes.Here, the dissociation is clearly dominated by a two-step process, suggesting that thermal energy is re-quired for maintaining a conformational changeneeded to stabilize the complex.

Strongly negative changes in heat capacity forthe ifnar1–IFNα2 complex formation

Analysis of the temperature-dependent equili-brium constants yielded a similar picture. For thebinary IFNα2/ifnar2-H10 complex, the equilibriumdissociation constant changedvery littlewith increas-ing temperature (Fig. 6a), yielding a linear correlationof ΔG0 with temperature (Fig. 6c) with a tempera-ture-independent ΔS0 of 63±9 J mol−1K−1 and ΔH0

of −27±3 kJ mol−1 (Table 3). For ligand binding toifnar1, however, the binding affinity increased sub-stantially when the temperature was increased from5 and 25 °C (Fig. 6b), resulting in a nonlinearcorrelation of ΔG0 with temperature (Fig. 6c). Thisresult indicates that substantial changes in heatcapacity ΔCp

0 upon complex formation cause ΔH0

and ΔS0 to be temperature-dependent. Therefore,this curve was fitted by an extended Gibbs–Helm-holtz equation,46 yielding a ΔH0 of 30±5 kJ mol−1

Fig. 5. Temperature dependence of the ligand binding kinetics. (a) Normalized dissociation curves for the interactionof IFNα2 with immobilized ifnar2-EC at different temperatures. (b) Normalized dissociation curves for the interaction ofIFNα2-HEQ with immobilized ifnar1-H10 at different temperatures. (c and d) Arrhenius plot for the association rateconstants (c) and the dissociation rate constants (d) for the interactions of IFNα2-HEQ with ifnar2-H10, ifnar1-H10 andH10-SD123.


and a ΔS0 of 230±20 J mol−1K−1 at 25 °C with achange in heat capacity of −6.7±0.7 kJ mol−1 K−1.The strong temperature dependencies of ΔH0 andΔS0 are shown in Fig. 6d, illustrating the transitionfrom an endothermic to an exothermic interactionbetween 25 and 37 °C. Such large negative changes inheat capacity upon complex formation have beenobserved mainly for protein–DNA complexes andhave been ascribed to large structural adaptationsupon complex formation.47 These results thus confirma substantial conformational rearrangement of ifnar1-EC upon ligand binding, accompanied by a large lossin hydrophobic surface area.48 From theΔS0 andΔCp

0

values, a rearrangement of ∼80 amino acid residuescan be estimated.47 Very similar results were obtainedfor IFNα2-HEQ binding to H10-SD123, suggestingthat the major rearrangement occurs in the threeN-terminal Ig-like domains of ifnar1-EC.

A multistep ligand-induced conformationalchange in ifnar1

We further explored the kinetics of the ligand-induced conformational change in ifnar1 by perform-ing stopped-flow experiments employing the fluores-

cence assays described above. AT655ifnar1-H10N349Cwas rapidly mixed with IFNα2 E58A at differentconcentrations while monitoring the increase in fluo-rescence. Since IFNα2 E58A binds ifnar1-EC withapproximately five times higher affinity than wild-type IFNα2 and since it can be produced in muchhigher yields than IFNα2-HEQ, IFNα2 E58A wasused for these experiments. The fluorescence inc-reased rapidly upon complex formation with verylittle deviation from amonoexponential function (Fig.7a). A strong increase in curvaturewas observedwithincreasing ligand concentrations. The fit of theconcentration-dependence of the apparent rate con-stants (Fig. 7c) yielded an association rate constant ofka=(9±2)×105 M−1 s−1 and a dissociation rate cons-tant of kd= (1.2±0.3) s−1. The dissociation rateconstant was confirmed by chasing AT655ifnar1-H10N349C in complex with IFNα2 E58Awith unlabeledifnar1-H10 (Fig. 7d), yielding a kd=(1.1±0.1) s−1.While both ka and kd are approximately five timeshigher than the rate constants determined by surface-sensitive detection, the equilibrium dissociation con-stant KD=1.3 μM is in good agreement with theKD=0.8 μM determined by surface-sensitive detec-tion (data not shown).

Fig. 6. Thermodynamic analysis of the complex formation. (a) Equilibrium binding signal normalized to themaximum signal obtained for the interaction of IFNα2 with immobilized ifnar2-EC at 10 and 25 °C. (b) Equilibriumbinding signals for the interaction of IFNα2-HEQ with immobilized ifnar1-H10 at 10 and 25 °C. (c) Temperaturedependence of ΔG0 for the interactions of IFNα2 with ifnar2-H10 fitted by linear regression and for the interactions ofIFNα2-HEQ with ifnar1-H10 and H10-SD123 fitted by the extended Gibbs–Helmholtz equation (details in theSupplementary Data). (d) Temperature dependence of ΔH0, ΔS0 and ΔG0 for IFNα2-HEQ binding to ifnar1-H10assuming a change in heat capacity ΔCp

0 of −6.7 kJ mol−1 as determined from the fit shown in (c).


The same set of experiments was carried out withOG488ifnar1-H10 N23C in complex with AT565tris-NTA in order to probe the kinetics of the rearrange-ment of the Ig-like domains (Fig. 7b). In this case, theincrease in fluorescence was not monoexponentialfor all ligand concentrations. In a control experimentwith OG488ifnar2-H10 in complex with AT565tris-NTAmixed with 20 μM IFNα2 E58A, no significantchange in fluorescence was observed (Fig. 7b), con-

Table 3. Thermodynamic parameters at 25 °C

Ifnar2-H10/IFNα2

Ifnar1-H10/IFNα2-HEQ

H10-SD123/IFNα2-HEQ

ΔG0 −46.5±0.3kJ mol−1

−38.1±0.2kJ mol−1

−38.3±0.2kJ mol−1

ΔH0 −27±3 kJ mol−1 30±5 kJ mol−1 30±4 kJ mol−1

ΔS0 63±9J mol−1K−1

230±20J mol−1K−1

230±20J mol−1K−1

ΔCp0 – −6.7 kJ mol−1 K−1 −7.4 kJ mol−1 K−1

EA (ka) 60±1 kJ mol−1 ∼25 kJ mol−1 ∼25 kJ mol−1

EA (kd) 85±2 kJ mol−1 ∼0 kJ mol−1 ∼0 kJ mol−1

firming the specificity of the signal characteristics.The curves were approximated over a time intervalof 120 s by a triexponential fit, in which two of thethree rate constants were fit globally for all concen-trations (Fig. 7a). The variable rate constant wasdependent on the concentration of IFNα2 E58Awitha slight deviation from a linear correlation (Fig. 7c).Two concentration-independent rate constants ofk2=0.3 s−1 and k3=0.03 s−1 were obtained from thefit. The relative signal amplitude corresponding to k2was 35–50% of the signal amplitude associated withk1 for all concentrations, supporting the significanceof this rate constant. The significance of the rateconstant k3 with a signal amplitude of ∼20% of thesignal amplitude associated with k1, however, isnot clear. The same behavior was observed forOG488ifnar1-H10 S147C, and very similar rate con-stants of the triexponential fit as well as similarrelative signal amplitudes were obtained (Fig. 7c).Similar multiexponential binding kinetics was ob-served for OG488ifnar1-H10 N286C, but the totalsignal amplitudes were too low for a detailed assess-ment of the rate constants. These experiments clearly

Fig. 7. Ligand-induced conformational change probed by stopped-flow fluorescence measurements. (a) Increase influorescence uponmixing 50 nM AT655ifnar1-H10with IFNα2 E58A at different concentrations (red, 1 μM; blue, 10 μM). (b)Increase in fluorescence upon mixing 50 nM OG488ifnar1-H10 N23C in complex with AT565tris-NTAwith IFNα2 E58A atdifferent concentrations (red, 1 μM; blue, 10 μM). As a control, the same experiment was carried out with OG488ifnar2-H10in complex with AT565tris-NTA with 10 μM IFNα2 E58A. (c) Comparison of- the concentration-dependent rate constantobtained for OG488ifnar1-H10 N23C, OG488ifnar1-H10 S147C and AT655ifnar1-H10. (d) Decrease in fluorescence uponmixing 25 nM AT655ifnar1-H10 preincubated with 1 μM IFNα2 E58Awith 10 μM unlabeled ifnar1-H10.


established a two- or multistep ligand-inducedconformational change in ifnar1-EC.

Discussion

The molecular mechanisms underlying signalactivation by cytokine receptors are currently underdebate. While the classic model is based on ligand-induced receptor assembly,49 the current view for anumber of class I cytokine receptors favors receptorpreassembly and subsequent activation by ligand-induced conformational changes.29,30,36,37,50,51 Alsofor the IFNγ receptor—a member of the class IIcytokine receptor family—preassembly and ligand-induced conformational changes have been demon-strated in live cells.35,52 These studies emphasizedthat the receptor conformation and conformationalchanges upon ligand binding play key roles in signalactivation. The architecture of the extracellulardomain of ifnar1 has puzzled researchers for severalyears until its important role in differential signalactivation emerged.39 While the three N-terminal

domains of ifnar1 are involved in ligand binding, theEMstructure of the ternary complex revealed that theC-terminal cytokine-binding module of ifnar1adopts a similar arrangement with respect to ifnar2as seen in other class I cytokine receptor complexes.27

Here, we could show that the simultaneous interac-tion of the three N-terminal domains of ifnar1 withthe ligand results in a substantial ligand-inducedrearrangement of the Ig-like domains of ifnar1. Usingsite-specific fluorescence labeling in situ through theC-terminal His tag, we demonstrate by intramole-cular FRET experiments that ligand binding inducesa stretching of ifnar1 by more than 10 Å. Using anelectron-transfer-based fluorescence assay, wefurthermore established that the ligand-induced con-formational changes are propagated to the mem-brane-proximal Ig domain of ifnar1, which does notinteract with the ligand. The key importance of thisdomain for receptor assembly and signaling haspreviously been shown by chimeras with the cor-responding domain of other class II cytokinereceptors. Propagation of conformational changeswas probably impaired in these chimeras, suggesting


that the ligand-induced conformational changes playa critical role for signal activation. Temperature-dependent binding studies confirmed large confor-mational rearrangements during ligand recognitionand furthermore indicated that complex formationoccurs in more than one step. This hypothesis wasverified by stopped-flow fluorescence experiments,which suggested that a further rearrangement of theIg-like domains takes place in a second step afterligand recognition. Taken together with the two-steprecruitment of the receptor subunits observed onsolid-supported membranes,38,53 a multistep assem-bly mechanism as schematically depicted in Fig. 8emerges.Interestingly, no differences in the conformational

changes in ifnar1 were observed upon binding ofIFNα2 and IFNβ, and EM averages revealed that thetwo ternary complexes also have very similararchitectures. While differential receptor assemblyby IFNs has been previously proposed to be res-ponsible for differential signal activation, our resultsdo not support this hypothesis. Even on the level ofrather small conformational changes in the mem-brane-proximal domains probed by photo-inducedelectron transfer, we could not detect differences inthe ifnar1 conformations in the ternary complexesformedwith IFNα2 and IFNβ. Recently, the differentstabilities of the interaction of IFNα2 and IFNβwithifnar1 have been shown to play a critical role for theirdifferential response patterns.39,40 This finding isinteresting in the context of the rather slow rateconstant (0.3 s−1) we have observed for the con-formational change after ligand binding. Since thelifetime of the ifnar1-IFNα2 interaction in the ternarycomplex on the membrane is in the same order ofmagnitude (kd=0.3 s−1),53 the probability for theconformational change to happen during a bindingevent is much lower than for IFNβ, for which thelifetime of the complex with ifnar1 is 40 times longer.If the conformational change is important for signalactivation, this difference between IFNα2 and IFNβcould be responsible for a higher potency of IFNβcompared to IFNα2 observed under conditions ofprolonged stimulation.

Fig. 8. Ligand-induced assembly of the type I interferon recto the high-affinity subunit ifnar2, ifnar1 is recruited to compligand-induced conformational change, which involves a maj

The exact structural consequences of the ligand-induced conformational changes remain to beelucidated. However, we have identified a trypto-phan residue close to the membrane (W347) with astrong loss in accessibility upon ligand binding,which is probably due to a switch between anexposed to a buried state upon ligand binding. Theseresults are to some extent consistent with a genericmechanism, which has recently been proposed forligand-induced activation of class I cytokine re-ceptors.28 This mechanism is based on a ligand-induced rearrangement of the WSXWS motif, whichis found in the class I cytokine receptor family.Weidemann et al.28 postulated that the accessibilityof the tryptophan residues is changed by a ligand-induced conformational change, thus modulatingthe interaction of the receptor ectodomains with themembrane. While ifnar1, as a member of the class IIcytokine receptor family, does not contain the above-mentioned WSXWS motif, it is conceivable thatligand binding also modulates the interaction withthe membrane in this receptor. Strikingly, the res-pective tryptophan residue is conserved throughoutthe low-affinity subunits of the class II cytokinereceptor subunits (IL-10R2, IFN-γR2 and IL-20R2),but not in the high-affinity subunits with the singleexception of IFN-λR1.Taken together, our results provide a first glimpse

into the intricate structural reorganization duringthe ligand-induced assembly of the type I interferonreceptor. It is very likely that these conformationalchanges are essential for signal activation, high-lighting the fine-balanced molecular mechanismsthat are involved in cytokine receptor signaling.

Materials and Methods

Materials

FEW647tris-NTA and AT565tris-NTA conjugates weresynthesized and stoichiometrically loaded with Ni(II)ions as described previously.41 ATTO 655 maleimide waspurchased from ATTO-Tec (Siegen, Germany), OG488

eptor on the plasmamembrane. After binding of the ligandlete the ternary complex on the membrane, followed by aor rearrangement of the Ig-like domains of ifnar1.


maleimide from Invitrogen (Karlsruhe, Germany). Theextracellular domains of ifnar1-H10, ifnar1-H10 mutantsand a fragment containing the threeN-terminal domains ofifnar1 fused to an N-terminal decahistidine tag (H10-SD123) were expressed in Sf9 insect cells using baculovirusinfection and purified by immobilized metal-chelatingchromatography as previously described.17,23 Pooledfractions of ifnar1-H10 were incubated with 1 mMethylenediaminetetraacetic acid (EDTA) on ice and furtherpurified by size-exclusion chromatography (Superdex 200,16-60; 20 mMTris, pH 8, 200 mM sodium chloride). Pooledfractions of the cysteine mutants were incubated with1 mM EDTA and the free cysteines were subsequentlyreduced by incubation with 100 μM dithiothreitol (for theifnar1-H10 mutants S147C, N286C and N349C) or 3 mMtricarboxyethylphosphine (for ifnar1-H10 N23C) for 1 h onice. The reducing agent and misfolded/aggregated pro-teins were removed by size-exclusion chromatography(Superdex 200, 20 mM Tris, pH 7.5, 50 mM sodiumchloride). A threefold molar excess of maleimide-functio-nalized fluorophore was added to the pooled proteinfractions and incubated overnight at 4 °C. The solutionwasdiluted fivefold with 5 mM Tris, pH 8.5, and loaded ontoan anion-exchange column (Sepharose Q, AmershamBiosciences). Nonlabeled and multilabeled species wereseparated by elution with a gradient of 0 to 500 mMsodium chloride in 5 mM Tris, pH 8.5. Prior to use, thelabeled proteins were run over a Superdex 200 size-exclusion chromatography column in Hepes-bufferedsaline (HBS; 20 mM Hepes, pH 7.5, 150 mM sodiumchloride). Typical labeling degrees were 0.5–0.7 fluoro-phores per ifnar1-H10 molecule for the purified proteins.In order to confirm the specificity of the labeling reactionunder these conditions, wild-type ifnar1-H10 was sub-jected to the same procedure, which did not yieldsignificant quantities of labeled protein after purification.Ifnar1-EC without a His tag was generated by proteolyticcleavage of an ifnar1-H10 variant carrying a factor Xacleavage site upstream of the C-terminal His tag. IFNα2and IFNα2-HEQ were expressed in Escherichia coli andpurified from inclusion bodies as described previously.54

Metal-binding contaminants were subsequently removedby affinity chromatography using aNi–NTA resin (Qiagen,Hilden, Germany). Ifnar2 with a C-terminal decahistidinetag (ifnar2-H10) and tag-less ifnar2-EC (ifnar2-tl) wereexpressed in E. coli and refolded from inclusion bodies asdescribed before.54

Fluorescence quenching assays in solution

Spectroscopic fluorescence measurements in solutionwere carried out with a Cary Eclipse (Varian) at 20 °C. In a120-μl cuvette (Hellma), 100 nM of fluorescence-labeledifnar1-EC was diluted in HBS supplemented with 1 mg/ml bovine serum albumin. OG488 was excited at 470 nmand the emission spectra were recorded between 490 to600 nm. Subsequently, 250 nM FEW647tris-NTA or AT565tris-NTA loaded with Ni2+ was added and carefully mixed,and the fluorescence spectra were recorded until thequenching was equilibrated (∼15 min). Subsequently, thebinding partner (1 μM IFNα2-HEQ or 500 nM IFNβ incomplex with ifnar2-tl) was added and carefully mixedand the spectra recorded. Complexes of IFNswith ifnar2-tlwere prepared by adding a 20% molar excess of ifnar2-tl.To compete out the ligand, 10 μM of unlabeled ifnar1-tlwas added, carefully mixed and the spectra recorded untilno further change in fluorescence was detectable.ATTO655-labeled proteins were excited at 640 nm and

the emission between 660 and 800 nm was recorded.Fluorescence lifetime measurements of ATTO655 wererecorded with a FluoTime 200 instrument (PicoquantGmbH, Berlin, Germany) using a 640-nm LED (PicoquantGmbH, Berlin, Germany) pulsed with a 200-ps pulsewidth and 20-MHz repetition rate for excitation.

Electron microscopy and image processing

Samples were prepared by conventional negativestaining with 0.75% (w/v) uranyl formate as describedpreviously.55 Images were collected with a Tecnai T12electron microscope equipped with a LaB6 filament andoperated at an acceleration voltage of 120 kV (FEI,Hillsboro, OR). Images were recorded on imaging platesat a magnification of 67,000× and a defocus value of−1.5 μm. Imaging plates were read out with a Ditabismicron imaging plate scanner (Ditabis Digital BiomedicalImaging System AG, Pforzheim, Germany) using a stepsize of 15 μm, a gain setting of 20,000 and a laser powersetting of 30%. Pixels (2×2) were averaged to yield a pixelsize of 4.2 Å on the specimen level. Particles were selectedinteractively from the images using BOXER, the displayprogram associated with the EMAN software package.56The selected particles (18,379 particles from 66 images forthe ternary complex formed with IFNα2-HEQ and 14,782particles from 112 images for the ternary complex formedwith IFNβ) were windowed into 64×64 pixel images. Theparticle images were rotationally and translationallyaligned and subjected to 10 cycles of multireference align-ment using the SPIDER software package.57 Each round ofmultireference alignment was followed by principalcomponent analysis and K-means classification specifying50 output classes. The references used for the first multi-reference alignment were randomly chosen from the rawimages.

Real-time solid-phase detection by TIRFS and RIf

Ligand binding to ifnar1-ECmutants was probed in realtime by using label-free RIf detection in combination withTIRFS. Thesemeasurements were performedwith a home-built setup as described earlier38 using the 488-nm line ofan argon ion laser for the excitation of OG488 and a 633-nmHe/Ne laser for excitation of ATTO655. Proteins weretethered onto solid-supported membranes containing 5%bis-NTA lipids,whichwere obtained by vesicle fusion ontothe silica substrate of the RIf transducer layer58 orimmobilized on a polyethylene glycol polymer brushfunctionalized with tris-NTA.59 All measurements werecarried out in HBS buffer as described in more detailbefore.17,23,38

Surface plasmon resonance

Label-free binding assays by surface plasmon resonancewere conducted with a BIAcore T100 (Biacore AB,Uppsala, Sweden). All measurements were carried out inHBS buffer supplemented with 0.05% surfactant P20.Ifnar1-H10, H10-SD123 and Ifnar2-H10 were immobilizedon different NTA-functionalized surfaces (either commer-cial Sensor Chip NTA from Biacore or SIA Kit Au Chipstreated with a bis-NTA thiol60). The ligand was injected ata flow rate of 30 μl/min. Binding to the immobilizedreceptor subunit was followed in real time. Binding toifnar1-H10 and H10-SD123 was probed by injectingIFNα2-HEQ. Binding to ifnar2-H10 was probed with


both wild-type IFNα2 and IFNα2-HEQ. Ligand-bindingassays were carried out at temperatures between 5 and40 °C with an increment of 5 °C. For each measurement,the chip was regenerated by sequential injections of500 mM imidazole in HBS, 200 mM EDTA in HBS, and20 mM NiSO4 in HBS and then reloaded with the His-tagged proteins for the next cycle. The association (ka) anddissociation rate constants (kd) were determined byindividual curve fitting with the BIAevaluation software(version 3.1) using a 1:1 binding model (monoexponential)with variable R0. Data analysis of temperature-dependentbinding assays was done according to establishedmethods46–48 as described in detail in the SupplementaryData.

Fluorescence correlation spectroscopy

Single-molecule experiments were carried out with ahome-built setup that was described previously.61 Theexcitation laser used was a pulsed diode laser at 640-nmwavelength (LDH 635, PicoQuant GmbH, Germany)generating pulses with ca 100-ps pulse width and40-MHz repetition rate. The laser light was sent througha single-mode glass fiber and subsequently collimated toform a beamwith Gaussian beam profile of ca 5-mm beamwaist radius. The beam was then focused through anapochromatic water-immersion objective (UPLSAPO 60×,N.A. 1.2 w, Olympus) into the sample solution. The sameobjective collected the fluorescence emission (epifluores-cence setup) that was separated from the excitation lightby a dichroic mirror (650DRLP, Omega Optical). A tubelens focused it onto a circular pinhole with 100-μmdiameter. After the pinhole, the light was split into twochannels and refocused onto two single-photon avalanchediodes (PDM-50-C, Micro Photon Devices) after passing aband-pass filter (690DF40, Omega Optical). Fast electro-nics (PicoHarp 300, PicoQuant) were used for recordingthe detected photons in time-correlated time-taggedrecording mode. From these raw data, the autocorrelationcurves were calculated by cross-correlating photons fromthe two different single-photon avalanche diodes. Bindingexperiments were carried out with 1 nM AT655ifnar1-H10mixed with 100 nM unlabelled ifnar1-H10 in HBScomplemented with 1 mg/ml bovine serum albumin.The complex with the ligand was formed by incubatingthe mixture with 1 μM IFNα2-HEQ for 5 min prior tomeasurement. Data were acquired over 30 min.

Stopped-flow fluorescence measurements

Stopped-flow measurements were carried out with anSF-61, DX2 Double Mixing Stopped-Flow System(Hi_Tech Scientific) at 26 °C in HBS, pH 7.5, supplementedwith 1 mg/ml bovine serum albumin. ATTO655-labeledifnar1-H10 N349C was excited with a He/Ne laser(633 nm), and emission was detected through a band-pass filter (640–750 nm). AT655ifnar1-H10 N349C (50 nM)was mixed with increasing concentrations of IFNα2 E58A(2, 8, 12.5 and 20 μM). Dissociation was measured bymixing 25 nM AT655ifnar1-H10 N349C preassociated with1 μM IFNα2 E58A with 10 μM unlabeled ifnar1-H10.OG488ifnar1-H10 mutants were excited with a xenon lampat 480 nm, and fluorescence was detected through a band-pass filter (500–570 nm). OG488 (50 nM)-labeled ifnar1-H10 mutants were preincubated with 100 nM AT565tris-NTA for 30 min and rapidly mixed with increasingconcentrations of IFNα2 E58A at a final concentration

between 1 and 10 μM. As a control, measurements werealso carried out for OG488ifnar2-H10 S35C as describedabove. Data analysis is described in detail in the Supple-mentary Data.

Acknowledgements

This project was supported by grants to J.P. fromthe DFG (PI 405/1 and SFB 628) and by the HumanFrontier Science Organization (RGP60/2002). J.P. isa Heisenberg professor funded by the DFG (PI 405/3) and was supported by the Cluster of Excellence“Macromolecular Complexes” at the Goethe Uni-versity Frankfurt (DFG Project EXC 115). We thankDr. Klaus Fendler and Juan Jose Garcia Celma forhelp with using their stopped-flow fluorescencespectrometer. The molecular EM facility at HarvardMedical School was established with a generousdonation from the Giovanni Armenise HarvardCenter for Structural Biology and is maintained byfunds from the National Institutes of Health,GM62580 (to T.W.).

Supplementary Data

Supplementary data associated with this articlecan be found, in the online version, at doi:10.1016/j.jmb.2008.01.017

References

1. Smith, P. L., Lombardi, G. & Foster, G. R. (2005). TypeI interferons and the innate immune response—more than just antiviral cytokines. Mol. Immunol. 42,869–877.

2. Stetson, D. B. & Medzhitov, R. (2006). Type Iinterferons in host defense. Immunity, 25, 373–381.

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

4. 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.

5. 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.

6. 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.

7. 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.

8. Rani, M. R., Foster, G. R., Leung, S., Leaman, D., Stark,G. R. & Ransohoff, R. M. (1996). Characterization of


beta-R1, a gene that is selectively induced by inter-feron beta (IFN-beta) compared with IFN-alpha.J. Biol. Chem. 271, 22878–22884.

9. 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 betaL sub-units of the type I interferon receptor. J. Biol. Chem.271, 23630–23633.

10. 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. 273,3144–31447.

11. 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.

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

13. 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.

14. 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.

15. 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.

16. 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.

17. 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.

18. Slutzki, M., Jaitin, D. A., Yehezkel, T. B. & Schreiber, G.(2006). Variations in the unstructured C-terminal tailof interferons contribute to differential receptor bind-ing and biological activity. J. Mol. Biol. 360, 1019–1030.

19. Quadt-Akabayov, S. R., Chill, J. H., Levy, R.,Kessler, N. & Anglister, J. (2006). Determinationof the human type I interferon receptor binding siteon human interferon-alpha2 by cross saturation andan NMR-based model of the complex. Protein Sci.15, 2656–2668.

20. Roisman, L. C., Jaitin, D., Baker, D. P. & Schreiber, G.(2005). Mutational analysis of the IFNAR1 binding siteon IFNalpha2 reveals the architecture of a weakligand-receptor binding-site. J. Mol. Biol. 353, 271–281.

21. 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.

22. Cutrone, E. C. & Langer, J. A. (1997). Contributions ofcloned type I interferon receptor subunits to differ-ential ligand binding. FEBS Lett. 404, 197–202.

23. Lamken, P., Gavutis, M., Peters, I., Van der Heyden,J., Uze, G. & Piehler, J. (2005). Functional carto-graphy of the extracellular domain of the type Iinterferon receptor subunit ifnar1. J. Mol. Biol. 350,476–488.

24. Jaks, E., Gavutis, M., Uzé, G., Martal, J. & Piehler, J.(2007). Differential receptor subunit affinities of type Iinterferons govern differential signal activation. J. Mol.Biol. 366, 525–539.

25. Cajean-Feroldi, C., Nosal, F., Nardeux, P. C., Gallet, X.,Guymarho, J., Baychelier, F. et al. (2004). Identificationof residues of the IFNAR1 chain of the type I humaninterferon receptor critical for ligand binding andbiological activity. Biochemistry, 43, 12498–12512.

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

27. Li, Z., Strunk, J. J., Lamken, P., Piehler, J. & Walz, T.(2008). The EM structure of a type I interferon–receptor complex reveals a novel mechanism for cyto-kine signaling. J. Mol. Biol. 377, 715–724.

28. Weidemann, T., Hofinger, S., Muller, K. & Auer, M.(2007). Beyond dimerization: a membrane-dependentactivation model for interleukin-4 receptor-mediatedsignalling. J. Mol. Biol. 366, 1365–1373.

29. Lu, X., Gross, A. W. & Lodish, H. F. (2006). Activeconformation of the erythropoietin receptor: randomand cysteine-scanning mutagenesis of the extracellu-lar juxtamembrane and transmembrane domains.J. Biol. Chem. 281, 7002–7011.

30. Brown, R. J., Adams, J. J., Pelekanos, R. A., Wan, Y.,McKinstry, W. J., Palethorpe, K. et al. (2005). Model forgrowth hormone receptor activation based on subunitrotation within a receptor dimer. Nat. Struct. Mol. Biol.12, 814–821.

31. Stroud, R. M. & Wells, J. A. (2004). Mechanisticdiversity of cytokine receptor signaling across cellmembranes. Sci. STKE, re7.

32. Grotzinger, J. (2002). Molecular mechanisms of cyto-kine receptor activation. Biochim. Biophys. Acta, 1592,215–223.

33. Frank, S. J. (2002). Receptor dimerization in GH anderythropoietin action—it takes two to tango, but how?Endocrinology, 143, 2–10.

34. Gent, J., Van Den Eijnden, M., Van Kerkhof, P. &Strous, G. J. (2003). Dimerization and signal transduc-tion of the growth hormone receptor. Mol. Endocrinol.17, 967–975.

35. Krause, C. D., Lavnikova, N., Xie, J., Mei, E.,Mirochnitchenko, O. V., Jia, Y. et al. (2006). Preassem-bly and ligand-induced restructuring of the chains ofthe IFN-gamma receptor complex: the roles of Jakkinases, Stat1 and the receptor chains. Cell Res. 16,55–69.

36. Constantinescu, S. N., Huang, L. J., Nam, H. & Lodish,H. F. (2001). The erythropoietin receptor cytosolic jux-tamembrane domain contains an essential, preciselyoriented, hydrophobic motif. Mol. Cell. 7, 377–385.

37. Kubatzky, K. F., Liu, W., Goldgraben, K., Simmerling,C., Smith, S. O. & Constantinescu, S. N. (2005). Struc-tural requirements of the extracellular to transmem-brane domain junction for erythropoietin receptorfunction. J. Biol. Chem. 280, 14844–14854.

38. Gavutis, M., Lata, S., Lamken, P., Müller, P. & Piehler,J. (2005). Lateral ligand–receptor interactions on mem-branes probed by simultaneous fluorescence-interfer-ence detection. Biophys. J. 88, 4289–4302.

39. Jaitin, D. A., Roisman, L. C., Jaks, E., Gavutis, M.,Piehler, J., Van der Heyden, J. et al. (2006). Inquiring


into 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.

40. Kalie, E., Jaitin, D. A., Abramovich, R. & Schreiber, G.(2007). An interferon alpha2 mutant optimized byphage display for IFNAR1 binding confers specifi-cally enhanced antitumor activities. J. Biol. Chem. 282,11602–11611.

41. Lata, S., Gavutis, M., Tampé, R. & Piehler, J. (2006).Specific and stable fluorescence labeling of histidine-tagged proteins for dissecting multi-protein complexformation. J. Am. Chem. Soc. 128, 2365–2372.

42. Lata, S., Reichel, A., Brock, R., Tampé, R. & Piehler, J.(2005). High-affinity adaptors for switchable recogni-tion of histidine-tagged proteins. J. Am. Chem. Soc. 127,10205–10215.

43. Neuweiler, H., Schulz, A., Bohmer, M., Enderlein, J.& Sauer, M. (2003). Measurement of submicrose-cond intramolecular contact formation in peptidesat the single-molecule level. J. Am. Chem. Soc. 125,5324–5330.

44. Oliveberg,M., Tan, Y. J. & Fersht, A. R. (1995).Negativeactivation enthalpies in the kinetics of protein folding.Proc. Natl Acad. Sci. USA, 92, 8926–8929.

45. Saecker, R. M., Tsodikov, O. V., McQuade, K. L.,Schlax, P. E., Jr, Capp, M. W. & Record, M. T., Jr (2002).Kinetic studies and structural models of the associa-tion of E. coli sigma(70) RNA polymerase with thelambdaP(R) promoter: large scale conformationalchanges in forming the kinetically significant inter-mediates. J. Mol. Biol. 319, 649–671.

46. Boniface, J. J., Reich, Z., Lyons, D. S. & Davis, M. M.(1999). Thermodynamics of T cell receptor binding topeptide-MHC: evidence for a general mechanism ofmolecular scanning. Proc. Natl Acad. Sci. USA, 96,11446–11451.

47. Spolar, R. S. & Record, M. T., Jr (1994). Coupling oflocal folding to site-specific binding of proteins toDNA. Science, 263, 777–784.

48. Spolar, R. S., Ha, J. H. & Record, M. T., Jr. (1989).Hydrophobic effect in protein folding and other non-covalent processes involving proteins. Proc. Natl Acad.Sci. USA, 86, 8382–8385.

49. Cunningham, B. C., Ultsch, M., De Vos, A. M.,Mulkerrin, M. G., Clauser, K. R. & Wells, J. A. (1991).Dimerization of the extracellular domain of the humangrowth hormone receptor by a single hormone mole-cule. Science, 254, 821–825.

50. Staerk, J., Lacout, C., Sato, T., Smith, S. O., Vainchenker,W. & Constantinescu, S. N. (2006). An amphipathicmotif at the transmembrane-cytoplasmic junctionprevents autonomous activation of the thrombopoietinreceptor. Blood, 107, 1864–1871.

51. Yang, N., Wang, X., Jiang, J. & Frank, S. J. (2007). Roleof the growth hormone receptor transmembranedomain in receptor predimerization and GH-inducedactivation. Mol. Endocrinol.

52. Krause, C. D., Mei, E., Xie, J., Jia, Y., Bopp, M. A.,Hochstrasser, R. M. & Pestka, S. (2002). Seeing thelight: preassembly and ligand-induced changes of theinterferon gamma receptor complex in cells. Mol. Cell.Proteomics, 1, 805–815.

53. Gavutis, M., Jaks, E., Lamken, P. & Piehler, J. (2006).Determination of the 2-dimensional interaction rateconstants of a cytokine receptor complex. Biophys. J.90, 3345–3355.

54. 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.

55. Ohi, M., Li, Y., Cheng, Y. & Walz, T. (2004). Negativestaining and image classification—powerful tools inmodern electron microscopy. Biol. Proced. Online, 6,23–34.

56. Ludtke, S. J., Baldwin, P. R. & Chiu, W. (1999). EMAN:semiautomated software for high-resolution single-particle reconstructions. J. Struct. Biol. 128, 82–97.

57. Frank, J., Radermacher, M., Penczek, P., Zhu, J., Li, Y.,Ladjadj, M. & Leith, A. (1996). SPIDER and WEB:processing and visualization of images in 3D electronmicroscopy and related fields. J. Struct. Biol. 116,190–199.

58. Lata, S., Gavutis, M. & Piehler, J. (2006). Monitoringthe dynamics of ligand–receptor complexes on modelmembranes. J. Am. Chem. Soc. 128, 6–7.

59. Lata, S. & Piehler, J. (2005). Stable and functionalimmobilization of histidine-tagged proteins viamultivalent chelator head-groups on a molecularpoly(ethylene glycol) brush. Anal. Chem. 77, 1096–1105.

60. Tinazli, A., Tang, J., Valiokas, R., Picuric, S., Lata, S.,Piehler, J. et al. (2005). High-affinity chelator thiols forswitchable and oriented immobilization of histidine-tagged proteins: a generic platform for protein chiptechnologies. Chemistry, 11, 5249–5259.

61. Böhmer, M., Pampaloni, F., Wahl, M., Rahn, H. J.,Erdmann, R. & Enderlein, J. (2001). Time-resolvedconfocal scanning device for ultrasensitive fluores-cence detection. Rev. Sci. Instrum. 72, 4145–4152.

Documents

Ligand Binding Induces a Conformational Change in ifnar1 ... JMB2008.… · Ligand Binding Induces a Conformational Change in ifnar1 that Is Propagated to Its Membrane-Proximal Domain