Báo cáo khoa học: Functional analysis of cell-free-produced human endothelin B receptor reveals transmembrane segment 1 as an essential area for ET-1 binding and homodimer formation pptx

We have now analyzed the quality and functional folding of cell-free produced human endothelin type B receptor samples as an example of the rhodop-sin-type family of G-protein-coupled re

endothelin B receptor reveals transmembrane segment 1

as an essential area for ET-1 binding and homodimer

formation

Christian Klammt1, Ankita Srivastava2, Nora Eifler3, Friederike Junge1, Michael Beyermann4,

Daniel Schwarz1, Hartmut Michel2, Volker Doetsch1and Frank Bernhard1

1 Centre for Biomolecular Magnetic Resonance, Institute for Biophysical Chemistry, University of Frankfurt ⁄ Main, Germany

2 Max-Planck-Institute for Biophysics, Department of Molecular Membrane Biology, Frankfurt ⁄ Main, Germany

3 M.E Mueller Institute for Microscopy, Biocentre, University of Basel, Switzerland

4 Leibniz-Institute of Molecular Pharmacology, Department of Peptide Chemistry & Biochemistry, Berlin, Germany

Keywords

cell-free expression; detergent micelles;

endothelin-1 ligand-binding site; G-protein

coupled receptor; single-particle analysis

Correspondence

F Bernhard, Centre for Biomolecular

Magnetic Resonance, Institute for

Biophysical Chemistry, University of

Frankfurt ⁄ Main, Max-von-Laue-Str 9,

D-60438 Frankfurt⁄ Main, Germany

Fax: +49 69 798 29632

Tel: +49 69 798 29620

E-mail: fbern@bpc.uni-frankfurt.de

(Received 27 March 2007, revised 26 April

2007, accepted 27 April 2007)

doi:10.1111/j.1742-4658.2007.05854.x

The functional and structural characterization of G-protein-coupled recep-tors (GPCRs) still suffers from tremendous difficulties during sample preparation Cell-free expression has recently emerged as a promising alter-native approach for the synthesis of polytopic integral membrane proteins and, in particular, for the production of G-protein-coupled receptors We have now analyzed the quality and functional folding of cell-free produced human endothelin type B receptor samples as an example of the rhodop-sin-type family of G-protein-coupled receptors in correlation with different cell-free expression modes Human endothelin B receptor was cell-free pro-duced as a precipitate and subsequently solubilized in detergent, or was directly synthesized in micelles of various supplied mild detergents Purified cell-free-produced human endothelin B receptor samples were evaluated by single-particle analysis and by ligand-binding assays The soluble human endothelin B receptor produced is predominantly present as dimeric com-plexes without detectable aggregation, and the quality of the sample is very similar to that of the related rhodopsin isolated from natural sources The binding of human endothelin B receptor to its natural peptide ligand endo-thelin-1 is demonstrated by coelution, pull-down assays, and surface plas-mon resonance assays Systematic functional analysis of truncated human endothelin B receptor derivatives confined two key receptor functions to the membrane-localized part of human endothelin B receptor A 39 amino acid fragment spanning residues 93–131 and including the proposed trans-membrane segment 1 was identified as a central area involved in endo-thelin-1 binding as well as in human endothelin B receptor homo-oligomer formation Our approach represents an efficient expression technique for G-protein-coupled receptors such as human endothelin B receptor, and might provide a valuable tool for fast structural and functional characterizations

Abbreviations

bET-1, biotinylated endothelin-1; C, cytoplasmic; CECF, continuous exchange cell-free; cET-1, Cy3-labeled endothelin-1; CF, cell-free; CTD, C-terminal domain; E, extracellular; ETB, human endothelin B receptor; ET-1, endothelin-1; GPCR, G-protein-coupled receptor; LMPG, 1-myristoyl-2-hydroxy-sn-glycero-3-[phospho-rac-(1-glycerol)]; NTD, N-terminal domain; RM, reaction mixture; SPR, surface plasmon

resonance; TMS, transmembrane segment.

G-protein-coupled receptors (GPCRs) form a large

superfamily of membrane proteins, and their genes

comprise an estimated 1–5% of vertebrate genomes

They modulate the activity of specific targets such as

ion channels or enzymes via G-protein coupling, and

thus initiate intracellular signaling cascades in response

to a broad range of external signals [1,2] GPCRs have

a similar architecture, composed of seven

transmem-brane segments (TMS1–7) connected by three

extracel-lular (E1–3) and three cytoplasmic (C1–3) loops In

addition, GPCRs contain a more or less extended

N-terminal domain and C-terminal domain, which are

both often involved either in ligand binding or

G-pro-tein coupling

Owing to their key role in signal transduction in

eukaryotic cells, GPCRs are estimated to represent the

targets for more than 50% of modern pharmaceutical

drugs [3] Despite much investigation, the

high-resolu-tion structural evaluahigh-resolu-tion of GPCRs as a prerequisite

for directed drug design is so far still limited to the

naturally abundant phototransducer rhodopsin [4] As

for many other membrane proteins, the first bottleneck

in structural and functional characterization of GPCRs

is the production of sufficient amounts of protein

sam-ple Although considerable improvements have been

made, the overproduction of GPCRs in cellular

expres-sion systems based on bacterial, yeast or insect cells is

still complicated and often inefficient [5–11]

Continuous exchange cell-free (CECF) expression

systems based on Escherichia coli cell extracts have

recently been demonstrated to provide a new and

highly promising tool for the preparative-scale

produc-tion of membrane proteins [12–14] Besides the

elimin-ation of toxic effects upon membrane protein

overproduction, a unique advantage of CECF systems

is the possibility of directly producing soluble

mem-brane proteins in the presence of detergents

[12,13,15,16] This completely new strategy provides an

artificial hydrophobic environment that is able to

inter-act with membrane proteins during translation Protein

precipitation is prevented, and functional folding

path-ways can be facilitated [17,18] The protein–detergent

association is initiated by hydrophobic interactions,

and specific targeting or translocation systems are

therefore not necessary

The endothelin (ET) system is involved in many

physiologic processes, such as control of vascular tone,

neurotransmission, embryonic development, renal

function, and regulation of cell proliferation, and it

thus plays an important role in physiopathologic

disor-ders such as congestive heart failure, diabetes,

athero-sclerosis, and primary pulmonary hypertension [19–21]

The human ET receptor type B (ETB) is a prototypic

GPCR distributed among multiple endothelial cell types as well as in smooth muscle cells, where it trans-mits vasoactive effects by binding the 21-mer isopep-tides ET-1, ET-2, and ET-3 ETB has equally potent affinities for ET-1, ET-2 and ET-3, in contrast to the homologous ET receptor type A, which has a higher affinity for ET-1 and ET-2

We have established protocols for the high-level pro-duction of ETB and other GPCRs in an individual CECF system [22] The GPCRs can be synthesized as precipitates or in soluble form in micelles of selected detergents, and apart from small terminal peptide tags that facilitate detection and purification, no large fusion proteins are needed for expression and stabiliza-tion The functional folding of membrane proteins overproduced by the new cell-free (CF) approach is of primary interest, and we therefore further analyzed the quality of ETB samples obtained after CF production under different conditions Ligand-binding and oligo-mer formation studies demonstrated that CF-produced ETB is functionally folded when synthesized in the presence of Brij detergents, and single-particle analysis revealed nonaggregated proteins that predominantly form dimeric complexes On the basis of the functional

in vitro analysis of rationally designed terminal ETB truncations, we specified a core domain responsible for ET-1 binding as well as for receptor dimerization in a relatively small region centered on TMS1

Results

CF production of ETB CECF reaction protocols were essentially performed as previously described [22] Full-length ETBcHxand trun-cated derivatives were produced as translational fusions with the small 12 amino acid T7-tag at the N-terminus and a poly(His)10-tag at the C-terminus (Fig 1) To obtain the highest yields of the individual constructs, optimization of the Mg2+ and K+ concen-trations in the ranges 12–16 mm and 250–340 mm was critical Under optimized conditions, ETBcHxcould by synthesized in yields of up to 3 mg per mL of reaction mixture (RM), and it was separated as a prominent band of c 46 kDa by SDS⁄ PAGE (Fig 2) The full-length synthesis of ETBcHxwas verified by immunode-tection with antibodies directed against the N-terminal T7-tag and the C-terminal poly(His)10-tag, respectively (data not shown)

ETBcHxwas CF produced either as a soluble protein

in the presence of detergents or as a precipitate in the absence of detergents We analyzed the yield and sample quality of ETB synthesized in the presence of

digitonin and long-chain Brij derivatives, as these detergents have been most effective for the soluble expression of GPCRs [15,22] In the presence of 1% Brij78 and 1.5% Brij58, completely soluble ETBcHx with final yields of c 3 mg proteinÆ(mL RM))1 was produced With Brij35 (0.1%) and digitonin (0.4%), only 500 lg and 100 lg of soluble ETBcHx per mL of

RM were obtained, respectively, in addition to ETBcHx precipitate The soluble ETBcHxproduced was purified

in one step by Ni-chelate chromatography, and, on average, c 60% of the synthesized ETBcHxin the RM was recovered ETBcHx precipitate CF produced in the absence of detergents was completely solubilized in 1% 1-myristoyl-2-hydroxy-sn-glycero-3-[phospho-rac-(1-glycerol)] (LMPG) and purified by Ni-chelate chro-matography

Coelution of purified ETB with ET-1 The mode of CF expression, i.e expression as precipi-tate or as soluble protein, as well as the type of

Fig 1 Proposed secondary structure of ETB The proposed seven TMSs are illustrated Various tags used for the modification of CF-pro-duced ETBs are indicated Predicted sites for post-translational modifications are a glycosylation site at N59, a disulfide bridge connecting E1 and E2, and several palmitoylation sites at cysteine residues in the CTD The first (black) and last (gray) amino acid positions used for the construction of truncated ETB fragments are indicated.

Fig 2 PAGE analysis of CF-expressed full-length ETB on a 12%

SDS gel M, marker Lane 1: RM control Lane 2: supernatant of

ETBcHx expression in the presence of Brij58 Lane 3: supernatant

of ETBcHx expression in the presence of Brij78 Lane 4: ETBcHx

after Ni–chelate acid chromatography Lane 5: supernatant of

ETB Strep expression in the presence of Brij78 Lane 6: ETB Strep after

Strep-Tactin purification Lane 7: ETBcHxprecipitate after expression

without detergent Lane 8: ETB cHx precipitate solubilized in 1%

LMPG Lane 9: ETB cHx precipitate solubilized in 1% LMPG after

Ni–chelate acid chromatography ETB is marked by arrows.

detergent, could have a significant impact on the

fold-ing of ETBcHx into a functional conformation The

affinity for its natural peptide ligand ET-1 was

there-fore analyzed with purified ETBcHx samples produced

under different conditions Mixtures of the

Cy3-dye-labeled derivative Cy3-Cy3-dye-labeled endothelin-1 (cET-1)

with purified ETBcHx, either CF produced as

precipi-tate and solubilized in 1% LMPG or directly produced

as soluble protein in the presence of 1.5% Brij58, 1%

Brij78 or 0.4% digitonin, respectively, were separated

by gel filtration, and the elution fractions were analyzed

by taking advantage of the different absorbancies of the

two compounds (Fig 3) The 52 kDa ETBcHxelutes at a

retention volume of 1.6 mL, whereas the 21-mer cET-1

starts to elute at a volume of 2.1 mL Coelution of

cET-1 with ETBcHxtherefore indicates complex

forma-tion of the receptor with its ligand, giving evidence of a native protein conformation In contrast, CF-produced ETBcHxpresent in an unfolded or inactive conformation should result in the separation of the two compounds cET-1 was completely separated from ETBcHx sam-ples that were CF produced as precipitate and solubi-lized in LMPG, indicating that, despite solubilization, the receptor might not have adopted its native confor-mation In contrast, significant amounts of cET-1 co-eluted with ETBcHxsynthesized in the soluble mode of

CF expression in the presence of digitonin, Brij58 and Brij78 The highest apparent binding of cET-1 was obtained with protein CF expressed in the presence of Brij78 This expression condition was therefore chosen for further sample preparations of ETBcHx and its derivatives

Fig 3 Functional conformation of full-length ETB cHx ET-1 binding of ETB cHx analyzed by coelution Purified ETB samples produced CF at dif-ferent conditions were incubated with cET-1 for 3 h at 21 C, and subsequently analyzed on a Superose 6 PC 3.2 ⁄ 30 column The elution chromatograms show total protein absorption at 280 nm (solid line) and specific absorption of cET-1 at 550 nm (dashed line) The retention volume of ETBcHx is indicated by arrows (A) ETBcHx CF expressed as precipitate and resolubilized in 1% LMPG; (B) soluble ETBcHx expressed in the presence of 0.4% digitonin; (C) soluble ETB cHx expressed in the presence of 1.5% Brij58; (D) soluble ETB cHx expressed in the presence of 1% Brij78.

The percentage of ligand-binding receptor present in

purified ETBcHx samples obtained after soluble CF

expression in the presence of 1% Brij78 was

deter-mined by correlation of the molar ratio of complexed

cET-1 with the amount of supplied ETBcHx After

background subtraction, we estimated the amount of

ligand-binding ETBcHx present in samples obtained

under the described conditions as c 50% This value is

similar to that for ETB samples obtained after

conven-tional expression in insect cells

Single-particle analysis of CF-produced ETBcHx

The quality of CF-expressed and purified ETBcHx was

analyzed by negative stain electron microscopy

ETBcHx protein that was CF synthesized in the

pres-ence of Brij78 revealed evenly distributed particles with

no detectable signs of aggregation (Fig 4A) ETBcHx

synthesized under these conditions appears to be

predominantly dimeric, and the good quality of the

sample allowed further structural assessment using

single-particle analysis Five hundred side views were

reference-free aligned, classified, and averaged within

the classes (Fig 4A) ETBcHx side view averages

play a pair of rods with a length of 63–68 A˚ The

dis-tance between the centers of the rods corresponds to

35–38 A˚, and the rods are closely associated at one

end These values are in excellent agreement with the

dimensions observed for the rhodopsin dimer [23,24]

Single rods, which presumably represent side views of

dimers but could also be ETBcHxmonomers, represent

less than 10% of all particles In contrast, and in

agreement with the observed inability to bind cET-1,

ETBcHx produced as a precipitate and solubilized in

LMPG was found to be aggregated, and is therefore

most likely unfolded (Fig 4B)

Localization of the ET-1-binding site

In order to determine the ETB region that is essential

for binding of ET-1, a series of eight plasmids coding

for terminally truncated ETB fragments and

contain-ing different secondary structural elements were

con-structed (Table 1, Fig 1) All fragments could be

overproduced in amounts of at least 1 mgÆ(mL RM))1

in our CF system as soluble proteins in the presence of

1% Brij78 (Fig 5A) The fragments were purified after

expression in one step by Ni-chelate chromatography,

and the purity was evaluated by SDS⁄ PAGE analysis

(Fig 5B)

The ligand binding of ETBcHxand truncated

deriva-tives was characterized by pull-down assays of purified

proteins with immobilized biotinylated ET-1 (bET-1),

as described in Experimental procedures Fractions containing complexes of bET-1 with ETB derivatives were separated by SDS⁄ PAGE and blotted, and the proteins were identified by immunodetection with antibody to T7-tag (Fig 6) Only fragments containing TMS1-like full-length ETBcHx, ETB131 [N-terminal domain (NTD)-TMS1], ETB168 (NTD-TMS2), ETB203 (NTD-TMS3) and also the NTD-deleted fragment ETB93 (TMS1-TMS3) were detected in the eluted fractions, and formed complexes with bET-1

Accord-A

B

Fig 4 Single-particle analysis of CF-produced ETB Representative views of electronmicrographs of the negatively stained ETB full-length construct (A) ETB produced as soluble protein in the pres-ence of Brij78 The ETB sample appears to be nonaggregated, and particles are predominantly dimeric (black arrow); ETB monomers can be seen occasionally (white arrow) Side view class averages

of reference-free aligned ETB dimers are displayed in the gallery on the right (B) ETB produced as precipitate and solubilized in LMPG The sample is no longer monodisperse, but rather forms aggre-gates.

ingly, proteins devoid of TMS1-like ETB132

(TMS2-CTD) and ETB204 (TMS4-CTD) did not interact with

bET-1

Analysis of ETBcHx–ligand interaction by surface

plasmon resonance (SPR)

Although the coelution approach gives good evidence

for a ligand-binding activity of CF-produced ETB

samples, it is primarily not a quantitative assay SPR allows the sensitive detection and quantification of molecular interactions in real time We immobilized bET-1 on the streptavidin surface of the biosensor chip and analyzed the direct binding of functionally active ETBcHx ETBcHx solutions with increasing concentra-tions from 10 nm to 250 nm were loaded on the bET-1 chip, and binding kinetics were evaluated using biaevaluation 3.1 software In general, signals obtained from the Biacore assay were lower than expected for loading of ETBcHx as a relatively large analyte, and this effect is probably due to ligand occlu-sion by the detergent micelles Binding constants were therefore not determined by steady-state kinetics, but

Table 1 Structural characteristics of CF-produced ETB derivatives With the exception of ETBDT7, all proteins contain additionally an N-ter-minal T7-tag x, included; –, deleted; ⁄ , partially truncated.

Included domains

C-terminal tag

ETB Strep E27–S443 49.5 x x x x x x x x x x x x x x x Strep

A

B

Fig 5 PAGE analysis of CF-expressed ETB fragments on SDS

gels (A) CF expression of ETB fragments in the presence of Brij78;

0.7 lL of supernatant (S) and 2 lL of eluate (E) after Ni–chelate

chromatography were analyzed The overproduced ETB truncations

are indicated by arrows (B) Soluble CF-expressed, purified and

reconstituted ETB fragments Nine microliters of each sample was

analyzed on a 16.5% SDS gel 1, ETB93; 2, ETB131; 3, ETB168; 4,

ETB203, 5, ETB306; 6, ETB132; 7, ETB204; 8, ETB307; 9, ETBcHx.

Arrows indicate the ETB derivative monomers Putative oligomeric

forms are also visible.

Fig 6 Ligand binding of ETB derivatives Binding of ETB deriva-tives to bET-1 was analyzed by pull-down assays Bound proteins eluted from avidin matrix were separated on 16.5% SDS gels and detected by immunoblotting with antibody to T7-tag Arrows cate detected bET-1-interacting ETB fragments, and asterisks indi-cate expected positions of noninteracting ETB derivatives M, marker.

rather by association and dissociation rates, which

were fitted by using 1 : 1 Langmuir models The

deter-mined kdwas used for calculation of ka, and the

bind-ing constants KD were determined from kd⁄ ka We

determined the binding constant KD for binding of

ETBcHxto bET-1 as 6.2 ± 1.7· 10)9 (Fig 7) Similar

assays with the C-terminal-truncated derivatives

ETB131 and ETB93 revealed KD values of

(2.7 ± 1.9)· 10)8and (1.7 ± 0.5)· 10)8, respectively

Identification of TMS1 as an essential element

for ETB dimerization

Several GPCRs are known to form dimers that remain

stable even after SDS⁄ PAGE analysis Protein bands

corresponding to dimers or even higher oligomers of

full-length ETBcHxand of most of the truncated

deriv-atives are visible after separation of purified protein

samples by SDS⁄ PAGE (Fig 5A,B) In addition, our

single-particle studies provided strong evidence of

ETBcHx dimer formation We therefore attempted to

identify the structural elements responsible for ETB

oligomerization by analyzing heterodimer formation

between full-length ETBStrepand the various truncated

ETB fragments in two different pull-down assays

First, purified ETB fragments and full-length ETBStrep

were incubated at equimolar concentrations and then

loaded on Strep-Tactin columns In a second assay,

the full-length ETBStrep receptor was coexpressed

with the various truncated fragments in CF reactions,

and the RMs were then loaded on Strep-Tactin

col-umns In both assays, the interacting protein fragments

were identified after washing, elution and SDS⁄ PAGE

separation by immunoblotting with antibody to T7-tag

(Fig 8)

In the coexpression assays, the synthesis of

full-length ETBStrep and that of the corresponding ETB

fragment was always visible by immunoblotting

(Fig 8A) After loading of the RMs on Strep-Tactin

columns, the fragments ETB93(TMS1-TMS3), ETB131

(NTD-TMS1), ETB168 (NTD-TMS2), ETB203

(NTD-TMS3) and ETB306 (NTD-TMS5) were coeluted

together with ETBStrep, indicating an interaction of the

proteins However, fragment ETB132 (TMS2-CTD)

lacking the TMS1 region was not detectable in the

eluted fraction, and therefore seems not to interact

with ETBStrep After mixing of purified proteins, again

the fragments ETB131 (NTD-TMS1), ETB168

(NTD-TMS2), ETB203 (NTD-TMS3) and ETB306

(NTD-TMS5) were found to interact with ETBStrep,

whereas fragments lacking TMS1, such as ETB132

(TMS2-CTD) and ETB204 (TMS4-CTD), could not be

coeluted with full-length ETBStrep and were localized only in the flow-through of the Strep-Tactin column (Fig 8B)

A

B

C

Fig 7 SPR response curves for the interaction of immobilized bET-1 with full-length ETB, ETB 131 and ETB 93 (A) Interaction of ETB between 10 and 250 n M (B) Interaction of ETB131 between

200 and 1600 n M (C) Interaction of ETB93between 10 and 200 n M

The high-level production of GPCRs in conventional

in vivo systems such as E coli or Pichia pastoris cells

can be very difficult and inefficient Successful

approa-ches require the construction of large fusion proteins or

intensive optimization [11,25] In addition, a variety of

steps in conventional expression and purification

proto-cols, such as kinetics of membrane insertion, saturation

of biosynthetic translocation machinery, control of

pro-teolysis, growth conditions, and extraction of

recom-binant membrane proteins from cellular membranes,

are highly critical and need intensive optimization

We have established a fast and efficient protocol for

the high-level production of functionally folded human

ETB and other GPCRs that eliminates most of the

critical steps of conventional expression systems, as

membranes and living cells are no longer involved

Furthermore, the proteolysis of synthesized membrane proteins can easily be prevented by protease inhibitors N-terminal digestion of ETB, which is frequently observed upon in vivo expression, was not detectable

by CF expression [26] Also, terminal truncated deriva-tives of ETB, which are often very difficult to express

in vivo, due to proteolysis or translocation problems, can be produced at high levels in the CF system [27]

A unique advantage of CF expression systems is the possibility of inserting membrane proteins directly into detergent micelles upon translation The efficiency of this solubilization mode was nearly 100% in the case

of ETB, as no residual precipitate was detectable and expression levels were similar to those obtained in the absence of detergent The CF approach is very straightforward, and purified ETB protein in sufficient amounts for structural analysis can now be obtained

in less than 2 days It should also be mentioned that the production of labeled membrane proteins, even with complicated label combinations, is easily feasible

by CF expression without the need for extensive opti-mization screens and without any loss of productivity [28–30] The CF expression technique might therefore become applicable also for the production of other GPCRs In this regard, we have already produced the porcine vasopressin type 2 receptor and the rat corti-cotropin-releasing factor precursor receptor at high levels of several milligrams per milliliter of RM by using protocols very similar to that for ETB [15,17,22]

In addition, larger thioredoxin fusions of the human M2 muscarinic acetylcholine receptor, of the human

b2-adrenergic receptor and of the rat neurotensin receptor have been produced in CF systems in yields approaching 1 mg of protein [16]

As observed for production of the porcine vasopres-sin type 2 receptor [15,22], only the steroid detergent digitonin and several long-chain polyoxyethylene deriv-atives such as Bri35, Brij58 and Brij78 were suitable for the CF synthesis of soluble ETB in milligram amounts Detergents of the Brij family are extremely mild deter-gents, being unable to disintegrate membranes, and they are tolerated by the CF transcription⁄ translation machinery in amounts far exceeding 100· critical micellar concentration [15] The quality of synthesized receptor can vary with the type of detergent, and the highest apparent ET-1 binding was obtained with Brij78, with some lower activity being seen in digitonin and other Brij derivatives In cellular systems, it is also known that the binding activity and structural integrity

of GPCRs can be sensitive to the supplied detergents during solubilization [31] The extraction of active lig-and-free ETB from cell membranes was only possible with digitonin [27] Accordingly, specific detergents

A

B

Fig 8 Analysis of dimerization of truncated ETB fragments with

full-length ETBStrep containing a StrepII-tag Interacting proteins

eluted from Strep-Tactin spin columns were separated on 16.5%

SDS gels and immunoblotted with an antibody against the T7-tag.

(A) Interaction of ETBStrep and truncated ETB fragments after

coexpression in the CF system in the presence of 1% Brij78 S,

supernatant of the RM; E, corresponding eluted fractions from the

Strep-Tactin columns (B) In vitro interaction of purified ETB

frag-ments with purified ETBStrep Bound fractions or flow-throughs (F)

were analyzed 1, ETB 131 –ETB Strep ; 2, ETB 168 –ETB Strep ; 3, ETB 203 –

ETBStrep; 4, ETB306–ETBStrep; 5, ETB132–ETBStrep, flow-through; 6,

ETB132–ETBStrep; 7, ETB204–ETBStrep, flow-through; 8, ETB204

–ETB-Strep M, marker; dotted arrow, full-length ETB strep ; solid arrow,

truncated ETB fragments; gray arrow, putative ETB Strep –ETB xx

heterodimers.

were required for the functional folding of other

mem-brane proteins, such as the nucleoside transporter Tsx,

during CF expression [15] CF-produced precipitates of

Tsx as well as of ETB did not adopt functional

confor-mations upon solubilization An initial screen for

suit-able detergents is therefore most important for the

production of functionally folded membrane proteins

during CF expression in the soluble mode ETB is

known to become post-translationally modified by

palmitoylation, phosphorylation, and glycosylation

However, these modifications do not play a role in the

ligand-binding capacities of ETB [32], and they are

most likely absent in CF-produced ETB, resulting in

more homogeneous sample preparations that might be

even more suitable for crystallization studies On the

other hand, disulfide bridge formation is very likely to

occur in CF systems as long as no specific chaperones

are required [33]

SPR studies of GPCRs are generally difficult to

per-form, due to the intrinsic properties of these proteins

Hydrophobic environments are necessary, and the SPR

sensitivity level requires high receptor concentrations

on the biosensor surface in order to detect the binding

of low molecular weight ligands Therefore, only a few

SPR measurements with GPCRs have been successful

so far [31,34], but these reports have shown that

lig-ands can bind solubilized GPCRs even in lipid-free

environments and without the need for membrane

reconstitution Most recently, a modified assay that

employs the detergent-solubilized neurotensin receptor

as the analyte has been described [35], and we

success-fully applied this approach to the characterization of

ETB Interestingly, for both GPCRs, the amplitude of

the observed response was lower than might be

expec-ted if the relatively high mass of the receptor used as

analyte is considered Ligand occlusion by

immobiliza-tion on the sensor chip surface, as well as limited

access to the ligand-binding site of the receptor due to

the presence of detergent molecules, might account for

this effect Although there is still some potential for

the optimization of this technique, e.g by systematic

evaluation of sensor chip surfaces or of linker

struc-tures, the good correlation of the findings presented

here with the published results obtained with

neuroten-sin receptor indicate that the SPR technique could

become a promising tool for the optimization of

GPCR expression conditions, for the localization of

ligand-binding sites, and for the identification of

com-pounds with new properties that could be important

for the pharmaceutical industry

Human ETB forms a very tight complex with ET-1

that remains stable even in 2% SDS [36] ET-1 binds

with high affinity to purified ETB in Brij78 micelles, as

indicated by the determined KDof 6 nm, which is even lower than the value of 29 nm previously determined

by total internal reflection fluorescence spectrometry with linear fluorescent labeled ET-1 [22] ETB⁄ ET-1 dissociation constants determined in vivo in various cellular environments range between 40 pm and

300 pm [37–41] It is known that the ligand-binding kinetics of ETB in intact cells are different from those

in corresponding membrane preparations [42] In addi-tion, the interaction of ETB in vivo with other pro-teins, such as G-proteins or receptor activity-modifying proteins, might dramatically increase the affinity for distinct ligands [43] In this work, we determined the dissociation constant of pure ETB in the environment

of detergent micelles, and this is also the first analysis

of ETB by SPR measurement The different assay con-ditions, in addition to the use of a modified biotinyl-ated ET-1 derivative as a ligand, have therefore most likely resulted in modified binding kinetics

The localization of ligand-binding sites in ETB is still a subject of controversy Labeling of ETB with radioactive ET-1, followed by chemical crosslinking and trypsin-digest analysis, located the ET-1-binding domain between residues I85 in the NTD and Y200 in the second cytoplasmic loop C2 [44] In addition, dele-tions, mutations and the lack of glycosylation in the NTD were found to have no effect on ET-1 binding to ETB [27] Our direct in vitro analysis of purified N-ter-minally and C-terN-ter-minally truncated ETB derivatives confined the ET-1-binding site to a 39 amino acid area between P93 in the NTD and C131 in the first cyto-plasmic loop C1 These data are in agreement with the above-mentioned findings, and they further define TMS1 as a central determinant for ET-1 binding On the basis of chimeric ETB derivatives and binding of antagonists, Wada et al proposed a 60 amino acid area spanning I138-I197, and thus covering TMS2 and TMS3, as the ET-1-binding site [44] In addition, other regions, such as TMS5, have been proposed to be involved in ligand binding as evaluated by

photoaffini-ty labeling with ETB-specific agonists [45] This result might have been caused by side-effects of the crosslink approaches, different binding sites of the supplied antagonists, or conformational changes of the analyzed chimeric ETB derivatives We showed that ETB truncations devoid of TMS1 but still retaining TMS2 and TMS3 are not able to bind ET-1 in detectable amounts Nevertheless, the affinity of ETB93 and ETB131 for ET-1 was reduced by approximately one order of magnitude, indicating that other regions of ETB still might contribute to the ligand binding Evi-dence for several and partially overlapping binding sites

of ETB for different ligands has been documented [46]

Homo-oligomerization of rhodopsin-like GPCRs is

an increasingly recognized mechanism, and might

rep-resent an important platform for the modulation of

GPCR activities such as ligand binding, signaling or

trafficking [24,47–49] Even SDS-resistant dimerization

of b2-adrenergic receptor and vasopressin type 2

recep-tor has been reported [47], and SDS-resistant dimers of

CF-produced porcine vasopressin type 2 receptor have

also been detected [15] The ETB dimer bands

observed during our SDS⁄ PAGE analysis indicate a

similar stable association The first evidence of the

for-mation of ETB homodimer and also of its homolog

human endothelin A in vivo was recently obtained by

fluorescence resonance energy transfer analysis in

HEK293 cells [50] Interestingly, ETB dimer formation

in vivodid not depend on the presence of ET-1 This is

in accordance with our observed oligomerization of

CF-produced ETB in the absence of any ligand

Fur-thermore, ETB dimer formation is strongly supported

by single-particle analysis, and the bilobed structures

described are almost identical to that of rhodopsin [24]

and to those of the vasopressin type 2 receptor and

corticotropin-releasing factor receptor type 1 [22] By

analyzing truncated ETB derivatives, we confined the

site that was essential for dimer formation to the

TMS1 fragment, which was also identified as covering

the ET-1-binding site The two fragments ETB131 and

ETB93, which overlapped in that region, did still form

homodimers as well as heterodimers with full-length

ETB Our results therefore indicate that TMS1 is a

key area for two main functions of ETB: the binding

of ET-1 as one of the main natural peptide ligands,

and ETB dimerization This close colocalization raises

the question of whether dimer formation could

modu-late the ligand-binding activity of ETB

In summary, the presented work provides an

interesting alternative approach for the generation of

high-quality samples for the functional and structural

characterization of ETB and similar GPCRs Further

analysis of the identified ETB131and ETB93fragments

will help to identify residues involved in ligand binding

and dimerization, and they might even represent

suit-able targets for structural studies by high-resolution

NMR analysis

Experimental procedures

CF expression

Proteins were produced in CECF systems essentially as

pre-viously described [14,22] Analytical-scale reactions for the

optimization of reaction conditions were performed in

mic-rodialyzers (Spectrum Laboratories, Rancho Dominguez,

CA, USA) with a molecular mass cut-off of 25 kDa in an

RM volume of 70 lL with an RM⁄ feeding mixture ratio of

1 : 14 Preparative-scale reactions were carried out in dispo-dialyzers (Spectrum Laboratories) in an RM volume of

1 mL with an RM⁄ feeding mixture ratio of 1 : 17 The reaction was optimized for the concentrations of the ions

Mg2+(15 mm) and K+(290 mm) For soluble expression, detergent was supplied during the reaction at the following final concentrations: Brij35, 0.1%; Brij58, 1.5%; Brij78, 1%; and digitonin, 0.4%

Cloning procedures and protein analysis Coding regions of full-length ETB and its derivatives were amplified from cDNA by standard PCR techniques, and the fragments were inserted into the expression vector pET21a(+) (Merck Biosciences, Darmstadt, Germany) Additional codons for extended poly(His)10-tags or for StrepII-tags were inserted by the Quickchange procedure (Stratagene, La Jolla, CA, USA)

Protein separated on 12% or 16.5% (w⁄ v) Tris ⁄ gly-cine⁄ SDS gels were transferred to 0.45 lm Immobilon-P poly(vinylidene difluoride) membranes (Millipore, Eschborn, Germany) blocked for 1 h in blocking buffer containing

1· Tris-buffered saline, 7% skim milk powder (Fluka, Buchs, Switzerland), and 0.1% (w⁄ v) Triton X-100 Horse-radish peroxidase-conjugated T7-tag antibody (Merck Bio-sciences) was diluted 1 : 5000 and incubated for 1 h with the membrane Washed blots were analyzed by chemilumines-cence in a Lumi-Imager F1 (Roche Diagnostics, Penzberg, Germany) Protein concentrations were determined by the bicinchoninic acid assay (Sigma, Taufkirchen, Germany) Soluble fractions diluted 1 : 10 in column buffer (20 mm Tris, pH 8.0, 500 mm NaCl) were applied to 1 mL His-trapHP columns (GE Healthcare, Freiburg, Germany) equil-ibrated in column buffer with 0.1% Brij78 Chromatography was performed at a flow rate of 1 mLÆmin)1 with washing steps of six column volumes of column buffer supplemented with 10 mm, 20 mm and 50 mm imidazole, respectively, and bound protein was eluted with 375 mm imidazole ETBStrep was purified on Strep-Tactin Spin columns (IBA, Go¨ttingen, Germany) according to the manufacturer’s recommenda-tions Precipitates produced CF in the absence of detergent were suspended in 1% LMPG in 20 mm phosphate buffer (pH 7.0), in volumes equal to the RM volume Suspensions were incubated for 1 h at room temperature with gentle sha-king, and this was followed by centrifugation for 10 min at

20 000 g (using an Eppendorf table top centrifuge 5810) in order to remove residual precipitate

Ligand-binding analysis The Cy3 dye was attached at Lys9 of cET-1 Biotin was covalently attached to Cys1 of ET-1 and Lys-9, result-ing in bET-1 For coelution studies of ET-1 with ETB,