Epidermal Growth Factor (EGF) - The Journal of Biological Chemistry

T H EJOURNAL OF BIOLOGICAL CHEMISTRY (0 1991 by The American Society for Biochemistry andMolecular

Vol. 266, No. 21, Issue of July 25, pp. 13828-13833, 1991 Printed in U.S.A.

Biology, Inc.

Epidermal Growth Factor (EGF) Induces Oligomerization of Soluble, Extracellular, Ligand-binding Domain of EGF Receptor A LOW RESOLUTIONPROJECTIONSTRUCTUREOF

T H E LIGAND-BINDING DOMAIN* (Received for publication, July 16, 1990)

Irit Lax$, Alok K. Mitrag, Chris Raveran, DavidR. Hurwitzll, Menachem Rubinstein$,Axel UllrichII , Robert M. Stroudg, and JosephSchlessingerS From the $.Department of Pharmacology, New York University Medical Center, New York, New York 10016, 7Rorer Central Research, Incorporated, King of Prussia, Pennsylvania 19406,11 Max Planck Institut fur Biochemie, 8033 Martinsriedbei Munchen, Federal Republic of Germany, and the $Departmentof Biochemistry and Biophysics, Universityof California School of Medicine, S a n Francisco, California 94143-0448 ”

Ligand-induced oligomerization is a universal phe- ing in rapid autophosphorylation aswell as the phosphorylanomenon amonggrowth factor receptors. Although the tion of various cellular substrates (1-3). The kinase activity mechanism involvedis yet to be defined, much evidence of EGFR is essential for signal transduction as kinase-negaindicates that receptor oligomerization plays a crucial tive EGFR mutants fail to elicit any EGF response (4-8). It role in receptor activation and signal transduction. is well established that the bindingof EGF to EGFR causes Here we show that epidermal growth factor (EGF) is rapid receptoroligomerization both in uitro and in intactcells able to stimulate the oligomerization of a recombinant, (9-17). Receptor oligomerization plays an important role in soluble, extracellular ligand-binding domain of EGF the underlying mechanism involved in the EGF-induced acreceptor. Covalent cross-linking experiments, analysis recepby sodium dodecyl sulfate-gel electrophoresis, sizeex- tivation of the tyrosine kinase activity and subsequent clusion chromatography, andelectron microscopy tor autophosphorylation(14-19). However, the kinase activity by ligand-independent mechdemonstrate that receptor dimers, trimers and larger of EGFR can also be stimulated multimers are formed in response toEGF. This estab- anisms, which may not involve receptor oligomerization (20, 21). lishes thatreceptoroligomerization is anintrinsic Interesting questions remaining to be answered include: property of the extracellular ligand-binding domain of involved inEGF-inducedreceptor EGF receptor. Ligand-induced conformational change what is the mechanism oligomerization and what are the structural elements essential in the extracellular domain will stimulate receptorreceptor interactions. This may bring about the allo- for this process t o occur? As receptor oligomerization is steric changeinvolved in signal transduction from the intimately associated with kinase activation, the unraveling extracellular domain across the plasma membrane, re- of the mechanism of this process will have important implisulting in the activation of the cytoplasmic kinase do- cations concerning our understanding of the mechanism of main. Electron microscopic images of individual extra- action of a number of transmembrane signaling molecules cellular ligand-binding domains appear as clusters of with similar structural topologies. The mechanism of EGFR foursimilarly-sizedstain-excludingareasarranged oligomerization is particularly intriguing, as it has shown been around a central, relatively less stain-excluded area. that each receptormolecule has a single binding site for EGF This suggeststhat the extracellular ligand-binding do- andEGFbindsto asingle EGFR (22). ThereforeEGFmain is structurally composed of four separate doinduced receptor oligomerization must involve receptor-recepmains. tor interactions that are stabilized by ligand binding (9). This is in contrast to platelet-derived growth factor receptors in which receptor dimerization is mediated by the various dimeric and therefore bivalent forms of platelet-derived growth The cell membrane receptor for epidermal growth factor factor (23-27). (EGF)’ belongs to thefamily of growth factor receptorswhich In this report we show that the extracellular domain of possess intrinsic protein tyrosine kinase activity (1-3). The EGF receptor isendowed with the capacity toundergo EGFbinding of EGF to the extracellular ligand-binding domain of dependent oligomerization leadingto the appearance of recepthe receptor activates its cytoplasmic tyrosine kinase, result- tor dimers, trimers, and even higher oligomerization states. We describe the two-dimensional projection structure of the * This research was supported by Grant GM24485 from the Na- extracellular ligand-binding domain determined from singletionalInstitutes of Health(to R. M. S.) andgrantsfromRorer particle averaging of electron images of a preferred orientaCentral Research and the Human Frontier Science Program (to J. S.).The costs of publication of this article were defrayed in part by tion. In thisview the molecule displays an approximate4-fold ligandthe payment of page charges. This article must therefore be hereby stain-excluding area,suggesting that the extracellular marked “advertisement” in accordance with 18 U.S.C. Section 1734 binding portion of EGFR is composed of four separate dosolely to indicate this fact. mains.

’

The abbreviationsused are: EGF, epidermal growth factor; EGFR, epidermal growth factor receptor; sEGFR, soluble epidermal growth MATERIALSANDMETHODS factor receptor; SPA, scintillation proximity assay; Hepes, 4-(hydroxPreparation of EGF Receptor Construct-The construct encoding yethy1)piperazineethanesulfonic acid; PMSF, phenylmethylsulfonyl fluoride; SDS, sodium dodecyl sulfate;PAGE, polyacrylamide gel the soluble, extracellularligand-bindingdomain of EGF receptor (sEGFR)denotedHERXCD was prepared bydigestion of CVN electrophoresis; DSS, disuccinimidyl suberate; CHO, Chinese hamHERC (28-30) with SacI. A 3124-base pair fragment containing a ster ovary.

13828

EGF Oligomerizationof EGFR Ligand-binding Domain

13829

portion of HER was subcloned into M13 and mutated with primer 5'in a solution of 2 parts Dl9 (Eastman Kodak) to 1 part distilled GGGCCTAAGATCTAGTAAATCGCCACTGGG 3'. The last amino water. acid residue of sEGFR is Ser-621 (28). The mutated construct was Image Analyses-Alignment of the images of single particles was cut with BsmI to yield a1357-base pair fragment containing the carried outby correlation-function-based rotational and translational mutated region. This fragment replaced the corresponding fragment alignment using the SPIDER softwarepackage (34). Micrographs in CNV HER. Following transfection, selection, and amplification were digitized by scanning on a flatbed micro-densitometer (model withmethotrexate,theChinesehamster ovary (CHO) cells were 1010M, Perkin Elmer) with50-pm aperture and step size, correspondgrown in medium composed of Dulbecco's modified Eagle's medium/ ing to 4.5 A on the specimen, and displayed on a Parallax 1280 display F12 (1:l) containing5% dialyzed serum by immunoaffinity and anion processor (Parallax Graphics, Sunnyvale, CA). Separate image files exchange chromatographies. for the smallest isolated particles, interpreted as the monomer of Purification of sEGFR-Recombinant sEGFR was purified from sEGFR, were created. Examinationof the calculated one-dimensional conditioned medium in two steps. The first step was affinity chroangular autocorrelation functions for a total of 97 particles showed a matography with monoclonal anti-EGF receptor (mAbl08) immobi- predominance of pseudo-4-fold character in the images. Sixty such lized oncyanogen bromide-activatedSepharosebeads(Pharmacia particles that displayed peaks 90 f 20" apart in the calculated angular LKB Biotechnology Inc.). The bound sEGFRwas washed extensively autocorrelation functions were selected. The images were low-passed with 500 ml of 10 mM Hepes, pH 7.2, 500 mM NaCl, 5% glycerol, filtered to a limiting highest resolution of 26 A and aligned both followed by 500 ml of 10 mM Hepes, pH 7.2, 100 mM NaC1. The rotationally and translationally by correlation methods (35) in two sEGFR was eluted of the column using 250 ml of Actisep, p H 7.5 stages of three cycles each. In the first stage, the reference was a (Sterogene Bioseporation, Inc., Acadia, CA), and dialyzed for 24-48 particle that visually best displayed the 4-fold character, and the h against 10 mM Hepes, 100 mM NaCl, with two changes of buffer. images were rotationally aligned withrespect to this image by calcuAfter the elution buffer was dialyzed away, the sEGFR was concen- lating one-dimensional cross-correlationsfrom -180" to 180" around trated using an Amicon stir cell (350 ml) and an Amicon 76 YM30 rings a t chosen radii in the images. The optimal values of the radii membrane.The second step of purification wasaccomplishedby were decidedbased on the rotational alignmentof two typical images. using anion exchange chromatography (Mono Q). The bound sEGFR The rotated images were translationally aligned to the reference by was eluted from the Mono Q column with 0-200 mM NaCl gradient, applying shiftsthat maximized the cross-correlation functions. In the and the eluted fractions were analyzed on a 4-12% SDS-PAGE Tris/ second stage, the rotationally and translationally shifted imaged were glycine gradient gel (Novex, Encinitas, CA). The fractions which added and averaged together to produce an image that served as the contain the purified sEGFR were pooled and concentrated using reference and the alignment procedure repeated. This second alignAmicon Centricon 30. Purified sEGFR was stored at 4 "C. ment-refinement cycle was expected to reduce bias in the alignment "'I-EGF Binding Experiments Using Scintillation Promixity Assay because of the choice of the particular reference in the first step. A (SPA)-The binding affinity of sEGFR for Iz5I-EGFwas determined scale factor was applied prior to theaveraging that made theaverage by SPA (31). SPA beads, PVT-RI (Amersham Corp.), are fluoromi- density in each image the same. The final cumulative rotational and crospheres coated with a monoclonal antibody, R1, directed against translational shifts were applied to the unfilteredimages to arrive a t the receptor extracellular domain.R1 does not compete with EGF for the set of 60 mutually aligned images. These aligned images were binding to the receptor. In this assay, "'1-EGF bound to sEGFR again scaled by multiplyingwith a scale factor which made theaverage coupled to PVT-RI beads generatesluminescence while free '"1-EGF density in each image the same. The overall average of the aligned does not (31). Hence, using the SPA method, the amount of bound images of the 60 particles,thestandard deviationbetween these I2'II-EGFcan be determined withoutphysically separating bound from aligned images, and two subset averages of independent sets of 30 unbound radiolabeled ligand. randomly chosen aligned images, for quantifying variations in the Binding assays were performed by either native EGF dilution or images,werecalculated. The density values a(i, j ) for themap by a method in which "'I-EGF was directly applied. In the native describing the standard deviation in theoverall average was given by EGF dilution method PVT-RI beads (50 pl), sEGFR (6.5 nM), "'Ithe following equation. EGF (0.5 nM, 100 pCi/pg) and nonradioactive EGF ranging from 10 t o 500 nM were incubated in 200 p1of20mM Hepes (pH 7.3), 0.1% bovine serum albumin, overnight at room temperature. Nonspecific u(i,j) = bm(i,;)- P(i,j))' binding was determined in triplicate in the presence of excess of (1) N-1 p ~ )The . direct methodwas performed similarly except native EGF (2 that increasing concentrations (2.5-1600 nM) of '"I-EGF were em- where p"(i, j ) and p ( i , j ) for pixel (i, j ) are the densityvalues for the ployed. Nonspecific binding was determined a t each concentration of mth image and the averaged map, respectively, and N is the number "'1-EGF by diluting with 100-fold excessnative EGF. All points were of images averaged. The averaging was carried out after multiplying determined in duplicate. Luminescencewas determined using a Beck- the densities in each scaled-aligned image with a weight that was man LS7000 scintillation counter with the window set for full range equal to the correlationcoefficient for the array of pixel densities in detection of "H. Following conversion of bound luminescence counts the given image to those in thereference, that were contained within t o bound y counts, binding affinity was determined by Scatchard a circular mask of diameter 72 A. analysis. The Fourier transforms of the two subset averages of the indeCovalent Cross-linking Experiments-Purified sEGFR (5-20 p ~ ) pendent sets of 30 scaled-aligned images were used to arrive a t a was incubated with EGF (10-30 p ~ in) 20 mM Hepes, pH 7.5, 150 measure of the resolution of the overall average. For thispurpose, the mM NaC1, for 1 h at room temperature. The covalent cross-linking mean amplitude-weighted phaseresidual A0 for the resolution shell s agent disuccinimidyl suberate (DSS) was added in a final concentra- = SI to s = SI given by the following equation (36, 37) was calculated. tion of 0.25 mM for 30 more min. An aliquot (30 pl) was mixed with two volumes of sample buffer, boiled for 4min, and analyzedby SDSPAGE (4-12%). Another aliquot (10 pl) was mixed with Tris-HC1 (IF11 + IF?I).(8, - 82)' buffer, pH 7.5, to a final concentration of 0.5 M , left for 1 h and then AO(s,, s a ) = "="I (2) analyzed by size exclusion chromatography on a TSKG4000SW col(IF11 + IF2213 umn (0.7 X 60 cm, TOSOH Corp., Japan). > = s, Preparation of Samples forElectron Microscopy-A 5-pl sample (110 pg/ml protein concentration) was applied to a 400-mesh copper Here I F I I and 1F21 are the amplitudes and81and O2 the phases of the grid covered with parlodian andoverlaid with a carbon film that was Fouriertransformcomponents a t a given resolution for the two rendered hydrophilic by glow-discharge in air. The sample was al- averages. A0 = 0" for images that are in perfect agreement and equals lowed to settle for 2 min, washed with 2 drops of distilled water, in theory104" for two completely uncorrelated images. The resolution stained with uranyl formate preparedaccording to Williams (32) for a t which A8 = 45" indicates a resolution less than equal or to which 20 s, and finally blotted by touching the edge of the grid with filter there is significant agreementbetween the Fourier components. paper. The grids were air-dried and examined in an EM400 electron microscope (PhillipsElectronicInstruments,Eindhoven,Holland) RESULTSANDDISCUSSION using an accelerating voltage of 80 kV. Micrographs were recorded In order to apply quantitative biophysical analyses todefine (33) at 111,200 X magnification, calibrated by usingthe 173.5 A specificspacing of (010) planes in negatively stained images of beef liver the molecular interactions involved in ligand binding catalase, on Kodak film 4489 (Eastman Kodak), developed for 4 min ity of EGF receptor and the mechanism of the subsequent

d

7

c

Jr" c

c

EGF Oligomerization of EGFR Ligand-binding Domain

13830

activation, we have generated a recombinant, soluble, extracellularligand-binding domain of human EGFR (sEGFR). CHO cells were transfected with a mammalian expression vector, which directs the synthesis of the extracellular domain of human EGFR (29). Following selection with neomycin and amplification with methotrexate, the CHO cells produced and secreted into the medium supernatant approximately 4 mg/ liter of sEGFR. The recombinant sEGFR migrated on SDS gels with anapparent M , of 105,000 (Fig. 1). sEGFRis glycosylated and uponremoval of sugars withdeglycosylating enzymes a protein core of 68,000 daltons remains. sEGFR was purified using affinity chromatography with monoclonal antiEGF receptor antibodies followed by anion exchange (Mono Q ) chromatography (Fig. 1 and "Materials and Methods"). The binding affinity of sEGFR for ""I-EGF was determined using a novel SPA (see "Materials and Methods" and Ref. 31) andScatchardanalysis of thebindingdata (Fig. 2). The dissociation constant Kd of ""I-EGF toward sEGFR is approximately 2.5 X 10" M. This value is 3-5-fold higher than the Kc, of '""IEGF toward the detergent-solubilized EGFR (38). The binding of""I-EGF to recombinant sEGFR was specific, as another growth factor, acidic fibroblast growth factor, did notinterferewiththebinding of ""I-EGF to sEGFR. Moreover, the binding of ""I-EGF to sEGFR was inhibited by a monoclonal antibody against EGFR, mAb96, which specifically blocks the binding of ""I-EGF to native A

-

EGF

B

+ - -I++

--

DSS

A B C D E F

m I

'4

1

-I"

FIG. 3. Analysis of sEGFR oligomerization by SDS-PAGE following covalent cross-linking. A, purified sRGFR (20 pM) was incubated with EGF (30 p ~ for) 90 min a t room temperature in 20 mM Hepes buffer, pH 7.4, 150 mM NaCl and subsequently with DSS for 1 h. Samples were analyzed by SDS-gel electrophoresis and stained with Coomassie Blue. R, Covalent cross-linking of sEGFR with EGF was done as in A. For increased sensitivity similar SDSgel runs were stained with silverstain. These analyses show various oligomerization states: I , monomers: 11, dimers; I l l , trimers; and IV, tetramers.

B C

kDa

272K134K 67K45K i t 1 i MONOMER(1OZK)

116 a

84-

e

'CrU .

FIG. 1. Generation of recombinant, soluble,extracellular domain of EGF receptor (sEGFR) in CHO cells. Coomassie Rluestained preparation of affinity-purified sEGFR isolated from conditioned medium of transfected cells is shown. sEGFR was purified by affinitypurification usinganti-EGFRmonoclonal antibody (mAbl0R) and by anion exchange chromatography using a Mono Q fast pressure liquid chromatography column. Lanes R and C show peak fractions eluted from the Mono Q column using 0-250 mM NaCl gradient. Lone A shows molecular weight standards.

.s

1

FIG. 4. Analysis of sEGFR oligomerization by size exclusion chromatography. DSS-cross-linked complex of sEGFR and EGF was prepared (see Fig. 3 for details). Cross-linking was terminated by addition of Tris-HCI buffer and aliquot (10 PI) wasanalyzed by chromatography on a TSKG4000SW column. The column was preequilibrated and eluted with a 150 mM NaCI, 20 mM Hepes, pH 7.5, buffer at a flow rate of 0.5 ml/min. The column was calibrated with molecular weight markers as indicated.

r

N

zx

Ve(ml)

1 400

8

Flee (nM)

. L i

U

50

0.5

m

n 0

2

1

3

Bound (nM) FIG. 2. Binding of '""IEGF and Scatchard plot of binding results. Binding was determined by SPA. Each point represents the average of two determinations.

cell surface EGFR (data not shown). We have previously described experiments utilizing a soluble covalent cross-linking agent to demonstrate that EGF induces receptor dimerization in vitro and in living cells (17). Hence, a similar approach was used to determine whether sEGFR aggregates upon binding of EGF. EGF and sEGFR were incubated together and subsequently treated with the covalent cross-linking agent DSS in order to render EGFinduced oligomer formation irreversible. The productsof this reaction can thusbe analyzedby SDS-gel electrophoresis,size exclusion chromatography, and electron microscopy. Fig. 3 shows that in the absence of either EGF or DSS essentially

EGF Oligomerization of Ligand-binding EGFR Domain

a

13831

b

FIG. 5 . Morphologies of uranyl formate-stained sEGFR. a, images of sEGFR alone; h, images of sEGFR treated with EGF and DSS; c-e, galleries of images of sEGFR particles in various oligomerized states c, monomers; d, dimers; e, trimers and larger oligomers. All samples were stored at 4 "C for several days prior$o preparation of EM grids. Bar, 1000 A.

trimers in the presence of the covalent cross-linking agent. Dimers were also seen clearly in the size exclusion chromatographyundernondenaturingconditions (Fig. 4). SDSPAGE analysis of fractions corresponding to thedimer peak (apparent M , 207,000) gave a major band corresponding to band ZZ in Fig. 3 while monomers peak gave a band corresponding to band I. It is noteworthy that thecovalent crosslinking experiments underestimate the extent of EGF-induced oligomerization of sEGFR because of the low efficiency of the covalentcross-linkingreaction.Controlexperiments with either acidic fibroblast growth factor orbovine serum albumin indicated that EGF-induced oligomerization of sEGFR was EGF-specific. A consistent pictureemerged fromthe inspection of sEGFR morphologies visualized by electron microscopy. Fig. 5a shows that in the absenceof EGF, monomers and small amount of dimers of sEGFR were observed at sEGFR concentrationsof 1-10 pg/ml used in sample preparation for microscopy. Essentially similar distribution of particle sizes were observed (data not shown)when sEGFR were treated with DSS. In the same concentration range, however, in the presence of EGF FIG. 6. Aligned images of sEGFR monomers. A gallery of digitized images of 60 selected EGFR monomers, in the preferred and DSS, dimers, trimers, and larger multimers of sEGFR orientation after rotational and translational alignment (see "Mate- were observed to be the prevalent forms of receptors (Fig. 5b). rials and Methods"). This figure was created using the software While it isnecessary to use DSS in order to detectoligomerpackage PRISM (43) developed for the Parallax 1280 display procesization in SDS gels, it is possible to observe EGF-induced sor. receptor oligomers by electron microscopy without DSS treatment (data not shown). Evidently, the forces that hold the monomers of sEGFR were detected. However, EGF in the receptors together are strong enough to withstand staining presence of DSS induced covalently linked receptor dimer, trimer, and a small amount of tetramer formation. Addition and electron microscopy. Inthe images of the unliganded, recombinantsEGFR, of different concentrations of EGF to the reaction mixture indicated that EGF-inducedoligomerization was saturable a t particle sizes increase incrementally. Many of the smallest particles, interpreted as monomers, display a preferred orifull receptor occupancy. EGF-induced receptor dimers and trimers were already detected at an EGF concentration of entation as they lie on the support film and in the predomi1.25 pM, which occupies only 5-10% of the available receptors. nant view appear as clusters of four, similarly sized stainQuantitation with either immunoblotting analysis with anti- excluding areas arranged arounda central less stain-excluding of thefinal rotationally and EGFR antibodies or by density scanning of various receptor area. Fig. 6 showsagallery forms indicated that approximately 16 and 13% of sEGFR translationally aligned images of 60 selected EGFR particles. molecules, respectively,were trapped in the form of dimers or Fig. 7A, in the form of contour plot and gray-level display,

13832

EGF Oligomerizationof EGFR Ligand-binding Domain shows the overall average of these 60 aligned images after scaling. Similar illustrations for two independent averages of subsets of 30 randomly chosen particles are shown in Fig. 7, B and C. Fig. 7 0 shows the standard deviationassociated with each pixel element of the overall average. The recombinant sEGFR molecule in the preferred orientation has an overallfour-lobed shape with dimensions as indicated in Fig. 7A. This averaged image shows four peaks of stain-excluding areas. Thenearest neighbor peaksare arranged 82-109" apart around the center and are 15-18 A distant from each other. Fig. 8 is a plot, as afunction of resolution, of the calculated weighted phase residual(see "Materials and Methods")for the transformsof the averages of two independent subsets of images (Fig. 7, B and C). As can be seen from this figure, the weighted phase residual Teaches a value of 45" for Fourier terms corresponding 21.5 to A resolution and increaseswith increasing resolution. In the nomenclature of Henderson et al. (39), a phase residual of 45" corresponds to an IQ 7 (on a scale whereIQ = 1corresponds 14 to random to perfect phase correspondence and I& phasing) or toa figure of merit of -0.67. This indicates that higher resolution Fourier components in the images are increasingly contaminated by noise. The limiting resolvtion of the averaged map in Fig. 7A was chosen to be 21.5 A below which there is significant agreement between the phases of the transformsof the subset averages. In the two independent averages of subsets of images (Fig. 7, I? and C ) the four-lobet shape is preserved and the dimensions agree to within -2 A between the centers of the lobes. The differences between the two maps in Fig. 7, B and C, in terms of the variability of the positions and the strengthsof the peaks, e.g. peaks 1 and 3, together with the standard deviation for the overall average (Fig. 7 0 ) establish the expected level of variation in the density map of Fig. 7A. The variations arise from residual errors in the alignment, stain artifacts, and systematicdifferences due to possible tilting of molecules on the grid or to different conformational states in the structure. The four similar lobes suggest that sEGFR is composed of four separate, similarly sized domains. This is not inconsistent with the observation of four domains with internal sequence homology in the extracellular portion of EGFR (40-42). Thus we propose that the four homologous domains in the sequence constitute thefour main lobes in the averaged image of sEGFR (Fig. 7A). This view of the sEGFR could be like the view of the whole membrane-bound receptor projected onto the membrane plane; however, it also could be a side view of the particle viewed parallel to the membrane, since we have not yet imaged membrane-bound receptor. The shaded area in Fig. 7A representing the stain-excludipg mass in this predominant view encloses an area of 1900 A'. The recombinant sEGFR includes 621 amino acids (68 kDa) and oligosaccharides and migrates on SDS gels with an apparent molecular mass of 105 kDa. Assuming a normal partial specific volume for this glycosylated recombinant pfotein of about 0.78 ml/gm, the expected volume is 135,000 A', Thus the structure may consist of four domains about 71 A long parallel to this predominant viewing direction. However, preliminary images of baculovirus-produced sEGFR, which are less glycosylated (90 kDa), show particles of size very similar to those of the CHO-produced sEGFR, suggesting that the

-

m D

F-l

-

Average density = 1.615 (optical density units). Positionsof the four peaks and the dimensions ofsthe particle are indicated. The shaded area encloses an area of 1900 A'. R and C, averages of two independent FIG. 7. Averaged imageof sEGFR monomers in the predom- sets of 30 randomly chosen aligned images. The contour intervals are inant orientation.A, the average of 60 aligned EGFR images shown the same as in D.The map of the standard deviation at each pixel of in Fig. 6 after scaling. Thebold, higher density contours enclose stain- the overall averaged image overlaid with the averaged image. Average excluding mass and the dotted, lower density contoursrepresent stain. density = 0.035 (optical density units). I

I

EGF Oligomerizationof EGFR Ligand-binding Domain

13833

3. Ullrich, A., and Schlessinger, J. (1990) Cell 6 1 , 203-212 4. Honegger, A. M., Dull, T. J.,Felder, S., Van Obberghen, E., Bellot,F., Szapary, D., Schmidt, A,, and Schlessinger, J. (1987) Cell 51,199-209 5. Honegger, A. M., Szapary, D., Schmidt, A,, Lyall, R., Van Obberghen, E., Dull, T. J., Ullrich, A,, and Schlessinger, J. (1987) Mol. Cell. Biol. 7 , 4568-4571 6. Chen. S. W.. Lazar. S. C.. Poenie. M.. Tsien. R. Y.. Gill. G.. and Rosenfeld.

8. Moolenaar, W.. H., Bierman, A. J., Tilly, B. C., K.,Honegger, A. M., Ullrich, , 7 , 707-710 9. Schlessinger, J. (1988) Trends Biochem. Sci. 1 3 , 443-447 10. Schlessinger, J., Schechter, Y., Willingham, M. C., and Pastan, I. (1978) Proc. Natl. Acad. Sci. U. S. A . 7 5 , 2659-2663 Resolution in A" 11. Zidovetzki. R.. Yarden. Y.. Schlessincer. J.. and Jovin., T. (1981) Proc. Natl. . Acad. Sci. U . S. A . 78,6981-6985- ' ' FIG. 8. Variation of weighted phase residual as a function 12. Yarden, Y., and Schlessinger, J. (1985) in Growth Factors in Biology and of resolution. A plot of calculated phaseresidual computed from the Medicine, pp. 23-45, Pitman, London Fourier terms for the transforms of the averages of two independent 13. Fanger, B. D., Austin, K. S., Eearp, S. H., and Cidlowski, J. A. (1986) subsets of images (Fig. 7, B and C) as a function of resolution (see Biochemistry 2 5 , 6414-6420 "Materials and Methods"). The phase residual reaches a value of 45" 14. Yarden, Y., and Schlessinger, J. (1987) Biochemistry 2 6 , 1434-1442 Yarden, Y., and Schlessinger, J. (1987) Biochemistry 2 6 , 1443-1451 at 21.5 A resolution as indicated,which represents the chosen nominal 15. 16. Bani-Schnetzler, M., and Pilch,P. F. (1987) Proc. Natl. Acad. Sci. U. S. A. resolution of the averaged image (Fig. 7 A ) . 8 4 , 7832-7836 17. Cachet, C., Kashles, O., Chambaz, E. M., Borrello, I., King, C. R., and Schlessinger, J. (1988) J. Biol. Chem. 263,3290-3295 oligosaccharides may sequester the negative stain and so not 18. Honegger, A. M., Kris, R., Ullrich, A,, and Schlessinger, J. (1989) Proc. Natl. Acad. Sci. L! S. A. 86,925-929 contribute to stain-excludingmass. Tkus the stain-excluding 19. Honegger, A. M., Schmidt, A,, Ullrich,A,, and Schlessinger, J. (1990) Mol. Cell. Biol. 1 0 , 4035-4044 four-domain structure would be -44 A thick if protein were 20. Koland, J. G., and Cerione, E. A. (1988) J. Biol. Chem. 2 6 3 , 2230-2237 the only stain-excluding mass, surrounding by oligosaccha21. Northwood, I. C., and Davis, R. J. (1988) J. Biol. Chem. 2 6 3 , 7450-7453 rides. A quantitative analysisof tilted images (in progress) is 22. Weber, W., Bertics, P. J., and Gill, G. N. (1984) J. Biol. Chem. 259,1463114636 required to arrive at the three-dimensional shape. 23. Hart, C. E., Forstrom, J. W., Kelly, J. D., Seifert,R. K., Smith, R.A,, Ress, R., Murray, M. J., and Bowen-Pope, D. F. (1988) Science 2 4 0 , 1529On the basisof these experimentswe conclude that receptor 1531 oligomerization is an intrinsic property of the extracellular 24. Heldin, C. H., Backstrom, G., Ostman, A,, Mannacher,A,, Ronnstrand, L., Rubin, K., Nister, M., and Westermark, B. (1988) EMBO J. 7 , 1387domain of EGFR.Theextracellulardomain of EGFR is 1393 endowed with at least two functions: binding of EGF and 25. Seifert, R. J., Hart,C. E., Phillips, P. E., Forstrom, J. W., Ross, R., Murray, M. J., and Bowen-Pope, D. F. (1989) J. Biol. Chem. 284,8771-8778 oligomerization. The oligomerization observed with the extraC.-H.,Ernlund,A,,Rorsman, C., and Ronnstrand, L. (1989) J . cellular domain alone is likely to reflect a role of this domain 26. Heldin, Biol. Chem. 264,8905-8912 in the process of EGF-induced oligomerization of the native 27. Hammacher, A,, Mellstrom, K., Heldin, C. H., and Westermark, B. (1989) EMBO J. 8 , 2489-2495 receptor, observed both in vitro andin living cells. It is 28. Ullrich, A., Coussens, L., Hayflick, J. S., Dull, T. J., Gray, A,, Tam,A. W., Lee, J., Yarden, Y., Libermann, T. A., Schlessinger, J., Downward, J., noteworthy, however, that in response to EGF, intact and Mayes, E. L. V., Whittle, N., Waterfield, M. D., andSeehurg, P. H. nativeEGFreceptorsdonot formoligomers largerthan (1984) Nature 309,418-425 dimers (6-9). It is possible that for the intact EGFR (which 29. Livneh, E., Prywes, R., Kashles, O., Reiss, N., Sasson, I., Mory, Y., Ullrich, A., and Schlessinger, J. (1986) J. Biol. Chem. 261,12490-12497 containsinadditiontotheextracellulardomain, asingle 30. Greenfield, C., Hils, I., Waterfield, M. D., Federwisch, M., Wollmer, A,, transmembranedomainand a largecytoplasmic domain), Blundell, T. L., and McDonald, N. (1989) EMBO J. 8,4115-4123 31. Bosworth, N., and Towers, P. (1989) Nature 3 4 1 , 167-168 more restrictive and limited receptor-receptor interactions 32. Williams, R. C. (1981) J. Mol. Biol. 1 5 0 , 399-408 are allowed. Alternatively or in addition, constrained by as- 33. Unwin, P. N. (1975) J. Mol. Biol. 9 8 , 235-242 Frank, J., Shimkin, B., and Dowse, H. (1981) Ultramicroscopy 6 , 343-358 sociation, the cytoplasmic domain of EGFR may dictate cer- 34. 35. Frank, J., Verschoor, A., and Wagenknecht,T.(1985) in New Methodologies tain conformational constraints on the extracellular domain inStudiesofProteinConfiguration (Wu, T. T., ed)pp, 36-89, Van Nostrand Reinhold, New York inhibitingtheformation of oligomerslarger thandimers. 36. Unwin, P. N. T., and Klug, A. (1974) J. Mol. Biol. 8 7 , 641-656 These constraints could be released upon separation of the 37. Wagenknecht, T., Grassucci, R., and Frank, J. (1988) J. Mol. Biol. 1 9 9 , 137-147 extracellulardomainfromthetransmembraneandcyto38. Yarden, Y., Harari, I., and Schlessinger,J. (1985) J. Biol. Chem. 2 6 0 , 315319 plasmic domains. Moreover, on the basis of our finding that R., Baldwin, J. M., Ceska, T. A., Zemlin, F., Beckmann, E., the EGF-induced oligomerization is a property of the extra- 39. Henderson, and Downing, K. H. (1990) J . Mol. Biol. 213,899-929 cellular domain alone, we propose that the functionof EGFR 40. Lax. I.. Bureess. W. H.. Bellot. F.. Ullrich. A,. Schlessincer. J.., and Givol. D: (1988)"Moi Cell. siol. 8 , 1831-1834 begins with the association mediatedby EGF binding,which 41. Bajaj, M., Waterfield, M. D., Schlessinger, J., Taylor, W. R., and Blundel, does not require signaling across the membrane. T. (1987) Biochem. Biophys. Acta 916,220-226 42. Lax, I., Bellot, F., Hawk, R., Ullrich, A., Givol, D., and Schlessinger, J. (1989) EMBO J . 8,421-427 REFERENCES 43. Chen, H., Sedat, J., and Agard, D. (1990) in Handbook ofBiological Confocal 1. Hunter, T., and Cooper, J. A. (1985) Annu. Reu. Biochem. 54,897-930 Microscopy (Pawley, J., ed) pp. 141-150, Plenum Publishing Carp., New 2. Carpenter, G., and Cohen, S.(1990) J. Biol. Chem. 2 6 5 , 7709-7712 York ,

,

I

-

I

~

~~~

~

~

~~I