NATURE BIOTECHNOLOGY VOL 17 NOVEMBER 1999 http://biotech.nature.com 1105

G protein–coupled receptors (GPCRs) are a large and ubiquitous classof membrane receptors1 and are important pharmacological targets2.G protein–coupled receptors activate membrane-bound het-erotrimeric guanosine triphosphate (GTP)-binding proteins (G pro-teins). In their resting state, the α-subunits of G proteins bind guano-sine diphosphate (GDP). Binding of an agonist to the receptorinduces formation of a ternary complex of ligand, receptor, and Gprotein, followed by nucleotide exchange on the G protein, and finallydissociation of the GTP-bound α-subunit from the βγ-complex andthe receptor. Subsequently, both the α-subunit and the βγ-complexinteract with downstream members of the signal cascade3. G pro-tein–coupled receptors are generally studied by complex cell-basedassays monitoring indirect downstream events of the signal transduc-tion cascade. Molecular assays that directly report G protein activa-tion could bridge the gap between the limited information obtainedfrom simple binding assays and the complexity of cell-based systems.

Immobilization of biomolecules has proved to be a valuablestrategy for molecular assays (e.g., in combination with opticalevanescent field techniques4). Proteins are usually immobilizedcovalently by C- or N-reactive groups attached to the solid substratesurface5. However, this can result in random orientation or alteredstructure of the protein, either of which could prevent optimal inter-action with sample molecules. The goal of this work was to find amethod that (1) leads to functional, oriented immobilization ofGPCRs to surfaces and (2) allows solid-state patterning of GPCRs tofacilitate surface-sensitive detection. These patterns would includereceptor-free areas to be used as reference (“local referencing”),enhancing the sensitivity of the assays.

Bovine rhodopsin, the dim light receptor of rod outer segmentsof the retina3, was chosen as a representative GPCR. In its photoacti-

vatable state, rhodopsin bears a chromophore, 11-cis-retinal, that iscovalently linked to the receptor by a Schiff ’s base and acts as aninverse agonist3, inhibiting the basal activity of opsin (the apopro-tein). Photoisomerization converts 11-cis- to all-trans-retinal, a fullagonist, which in turn triggers the active conformation of the recep-tor. This receptor was selected because it can be purified easily andits function is well understood. Upon activation, transducin, the Gprotein interacting with rhodopsin, desorbs with its α-subunit fromthe membrane6. Previous work has shown that this can be assayed bysurface plasmon resonance (SPR) as a decrease in mass charge on thesensor chip7. Therefore, we chose SPR to follow immobilization ofthe receptor and its interaction with the G protein.

ResultsConcept for immobilization of GPCRs. A glycosylation consensussite close to the extracellular N terminus is conserved amongsequenced GPCRs, and glycosylation has been confirmed for sever-al receptors8. We exploited this feature by using carbohydrate-spe-cific chemistry9 for biotinylation, thereby confining the biotin tag tothe extracellular domain of the receptor (Fig. 1A). This resulted inuniform orientation upon immobilization, so that the intracellulardomain of the receptor pointed away from the surface, optimizingthe interaction with the G protein. The sensor chip was coveredwith a mixed self-assembled monolayer (SAM)10 consisting ofbiotinylated thiols11 and an excess of ω-hydroxy-undecanethiol(HTA), to which streptavidin was bound. Micropatterns of thereceptor were created using microcontact printing12 to pattern theSAM with respect to the biotinylated thiols, resulting in micropat-terns of streptavidin. The immobilized receptor was stable both forhours and during many activation cycles. Binding of 11-cis-retinal

RESEARCH

Micropatterned immobilization of a Gprotein–coupled receptor and direct

detection of G protein activationChristoph Bieri1, Oliver P. Ernst2, Stephan Heyse1, Klaus Peter Hofmann2, and Horst Vogel1*

1Swiss Federal Institute of Technology, Institute of Physical Chemistry, Laboratory for Physical Chemistry of Polymers and Membranes, CH-1015 Lausanne,Switzerland. 2Institut für Medizinische Physik und Biophysik, Universitätsklinikum Charité, Humboldt-Universität zu Berlin, Germany.

*Corresponding author (e-mail: Horst.Vogel@epfl.ch).

Received February 2, 1999; accepted August 5, 1999

G protein–coupled receptors (GPCRs) constitute an abundant family of membrane receptors of highpharmacological interest. Cell-based assays are the predominant means of assessing GPCR activation,but are limited by their inherent complexity. Functional molecular assays that directly and specificallyreport G protein activation by receptors could offer substantial advantages. We present an approach toimmobilize receptors stably and with defined orientation to substrates. By surface plasmon resonance(SPR), we were able to follow ligand binding, G protein activation, and receptor deactivation of a repre-sentative GPCR, bovine rhodopsin. Microcontact printing was used to produce micrometer-sized pat-terns with high contrast in receptor activity. These patterns can be used for local referencing to enhancethe sensitivity of chip-based assays. The immobilized receptor was stable both for hours and during sev-eral activation cycles. A ligand dose–response curve with the photoactivatable agonist 11-cis-retinalshowed a half-maximal signal at 120 nM. Our findings may be useful to develop novel assay formats forGPCRs based on receptor immobilization to solid supports, particularly to sensor surfaces.

Keywords: G protein–coupled receptors, chip-based functional real-time assays, micropatterning

© 1999 Nature America Inc. • http://biotech.nature.com©

199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://b

iote

ch.n

atu

re.c

om

to opsin and the photoconversion of bound 11-cis-to all-trans-reti-nal could readily be measured as functions of ligand concentrationand light intensity, respectively.

Patterned immobilization of the receptor. We first tested theimmobilization protocol on homogeneous, nonpatterned, mixed(biotin-thiol/HTA) SAMs. Streptavidin (170 nM) binding led to aresonance angle shift of 0.24 ± 0.018° (averages are from at least nineexperiments). After washing with buffer and detergent, detergent-solubilized biotinylated washed membranes were added to a finalconcentration of 1 µM rhodopsin, which induced a resonance angleshift of 0.181 ± 0.02°. In some cases, potential sites of nonspecificprotein adsorption were blocked before streptavidin binding by 10mg/ml bovine serum albumin (BSA), followed by thorough washing(>10 cuvette volumes) with 1 mg/ml BSA in buffer R. The BSAinduced only minor angle shifts.

Patterned SAMs consisting of alternating stripes (width = 200 µm)of pure HTA (denoted reference region) and bars of a mixed SAM(biotin-thiol/HTA; denoted sample region) were produced by micro-contact printing. The immobilization steps were followed simultane-ously on the two regions by a home-built SPR setup that allows formeasurements that are both time resolved and laterally resolved7.Binding of streptavidin and biotinylated receptor to the sample regionwas indistinguishable from binding to homogeneous mixed SAMs(Fig. 2A and B). However, streptavidin binding to the reference regionwas negligible, and binding of biotinylated receptor (because of non-specific adsorption) was 10 times lower. The patterned immobiliza-tion of rhodopsin was further supported by an antibody directedagainst the nine C-terminal (intracellular) amino acids of rhodopsin,which specifically bound to the sample region (Fig. 2C).

After immobilization of the receptor and washing with detergent,a supported lipid bilayer (SLB)13 was formed by micellar dilution14,followed by addition of the G protein. G protein binding to the SLBwas fast (k = 0.17 s-1, monoexponential fit) and showed a half-maxi-

mal signal at 1.1 µM (not shown), comparable to previous results7. Itamounted to 0.24 ± 0.016° at a concentration of 2.5 µM. We couldnot detect a significant difference in total binding of the G protein tohomogeneous SLBs or the two regions on patterned SLBs, as theonly difference between the two regions is their receptor content.

Activity of the immobilized receptor. When GTP was added tothe system, no resonance angle shift was observed, indicating thatthe immobilized receptor was not constitutively active, consistentwith its native state. However, a light flash induced a fast decrease ofthe resonance angle (Fig. 2D), resulting from the activation of the Gprotein and its desorption from the membrane (Fig. 1C). We con-sider this desorption signal to be direct evidence of the interactionbetween the immobilized receptor and its G protein. Moreover,under the chosen experimental conditions, the initial slope of theobserved desorption signal, klin, is a good approximation for thereceptor intensity per unit area of surface. The values for klin were0.06 ± 0.015° per second (five experiments) using identical condi-tions. The activity in the sample region of patterned chips was simi-lar, whereas it was 25 times lower in the reference region (Fig. 2D).

Upon activation, cleavage of the Schiff ’s base between receptorand chromophore leads to the slow and spontaneous decay ofrhodopsin to opsin, the apoprotein, and all-trans-retinal15. The acti-vated G protein also relaxes to its resting state by hydrolysis of thebound GTP even faster (1/k = 20 s. for GTP hydrolysis in isolated Gproteins16). With time, therefore, the system moves to a state con-taining opsin and resting, membrane-bound G protein (Fig. 1D).This relaxation was observed as a slow (k = 0.003–0.006 s-1, monoex-ponential fit) increase of the resonance angle, which finally returnedto close to the starting value (Fig. 3A).

Ligand binding to the receptor. Opsin can be considered a GPCRthat binds inverse agonists (11-cis- or 9-cis-retinal), which in turn canbe converted to full agonists by photons. In this view, the opsin/retinalsystem behaves as a ligand-activated GPCR. To test for ligand binding,

1106 NATURE BIOTECHNOLOGY VOL 17 NOVEMBER 1999 http://biotech.nature.com

RESEARCH

Figure 1. (A) Immobilization of the receptor.Streptavidin binds to biotin-thiols in a mixedSAM and docks biotinylated receptor to thesurface. Carbohydrate-specific biotinylationof the latter results in uniform orientation ofthe bound receptor. The G protein binds tothe supported lipid bilayer, which is formedafter receptor immobilization. (B)Photoisomerization of 11-cis- to all-trans-retinal triggers the active conformation ofrhodopsin. G protein bound to rhodopsinliberates its GDP and (C) desorbs from themembrane upon GTP binding, leading to adecrease in mass charge on the sensor chipthat can be followed by SPR. (D) Activerhodopsin decays spontaneously to all-trans-retinal and opsin, and the G proteinbinds upon nucleotide hydrolysis againto the surface. (E) 11-cis- and 9-cis-retinalboth bind to opsin, regenerating photo-activatable rhodopsin.

Figure 2. Immobilization andreceptor activity on a micro-patterned SAM was followed bytime-resolved and laterallyresolved SPR on both regionssimultaneously. (A) Streptavidin,(B) biotinylated, detergent-solubilized washed membranes,and (C) antirhodopsin antibodyspecifically bound to the sampleregion (G), and only to a smallextent to the reference region(P). (D) The activity of thereference region amounts toonly 4% of the activity of thesample regions.

A C DB

A B C D E

© 1999 Nature America Inc. • http://biotech.nature.com©

199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://b

iote

ch.n

atu

re.c

om

NATURE BIOTECHNOLOGY VOL 17 NOVEMBER 1999 http://biotech.nature.com 1107

immobilized rhodopsin was almost completely converted to opsin byrepeated flashes. Then, either 11-cis- or 9-cis-retinal was added andallowed to bind to the receptor (Fig. 1E). The desorption signal uponphotoactivation revealed that 60% or 90% of the initial activity wasregained. (The difference between the two ligands reflects the differ-ent absorption spectra of the generated rhodopsin isoforms ratherthan different ligand binding efficiencies). Based on this result, wehypothesized that an excess of chromophore over receptor should leadto a system in which the activity is constant after each flash, because allgenerated opsin is reconverted to rhodopsin. Indeed, the samerhodopsin activity was measured during long periods of time (up to 6h) and during several activation cycles (Fig. 3A) at sufficient concen-trations of 11-cis- or 9-cis-retinal (10 µM). We consider the stability ofthe reconstituted receptor G protein system striking.

To quantify the opsin–retinal interaction, we first assessed theresponse of klin on different flash intensities (i.e., on different con-centrations of activated rhodopsin). At a constant concentration (10µM) of 11-cis-retinal, but increasing light intensities of the flashes (λ≥ 495 nm), a dose–response curve was obtained (Fig. 3C) that satu-rated at approximately 50% of the maximal possible flash intensity(which was limited by the instrumentation). Given that at this con-centration of 11-cis-retinal all opsins usually contained bound lig-and, we concluded that light flashes of ≤50% intensity were mostsensitive to changes of ligand binding by opsin.

Therefore, we used flashes of 50% intensity to assess the concen-tration-dependency of the 11-cis-retinal binding to opsin. Eachtime, rhodopsin was completely photolysed before addition ofincreasing amounts of 11-cis-retinal, and the activity of the formedrhodopsin was measured by a flash after 40 min (Fig. 4). The result-ing dose–response curve revealed a half-maximal concentration of

120 nM and saturation point close to the level of activity that wasmeasured before the first photolysation.

DiscussionWe have used carbohydrate-specific chemistry for the oriented andpatterned functional immobilization of a representative GPCR. Thereconstituted receptor–G protein system was remarkably stable, andrepeated activation/deactivation cycles could be monitored readily.The high contrast in patterning of the receptor activity is promisingfor future applications of chip-based assays, as they allow for local ref-erencing to enhance sensitivity and stability. A major advantage of thesystem is that G protein activation can be followed directly. Recently,the interaction between G proteins and their receptors was suggestedas a new pharmacological target17. For rhodopsin and transducin,inhibition of interaction by a small synthetic peptide has been report-ed18. The experimental procedure presented here is well suited to testother molecules for effects on G protein–receptor interactions.

The finding that immobilized GPCRs remain functional was notobvious a priori, as GPCR activation is thought to involve subtle con-formational changes3 that seemed likely to be hindered by immobi-lization. Potential future applications of this finding include immobi-lization-based, functional assays for GPCRs, particularly inflowthrough assay formats where different compounds (e.g., a library)may be tested serially with the same receptor preparation. Detection ofG protein activation may be carried out in bulk (e.g., by fluorescencespectroscopy and using suitable tagging), but more promising are sur-face-sensitive techniques such as SPR or total internal reflection spec-troscopy because of their excellent discrimination between bulk(unspecific) and surface (e.g., receptor-induced) effects. AlthoughSPR is useful to study G protein–receptor interaction, it is not wellsuited to detect binding of small molecules (such as ligands) directlybecause it is sensitive only to changes in mass charge. Total internalreflection spectroscopy has been shown to be a valuable tool to studymembrane proteins19 and has much higher sensitivity than SPR.Fluorescence tagging or fluorescent GTP analogs20 may be employedto detect ligand binding and G protein activation in those assay for-mats. Current research in our group focuses on the development ofsuch immobilization-based assays for hormone-activated GPCRs.

Experimental protocolMaterials. Buffer R: 0.15 M NaCl, 1 mM MgCl2, 10 mM sodium phosphate,pH 7; Buffer G: 0.13 M NaCl, 1 mM MgCl2, 1 mM dithiothreitol (DTT), 20mM Tris pH 7.1 (DTT was added just before use). Buffer ROG50 was buffer Rsupplemented with 50 mM octylglucoside (Alexis Biochemical Corp., SanDiego, CA). All chemicals were reagent grade or better. ω-hydroxy-unde-canethiol (HTA) was synthesized using standard protocols. We produced 12-

RESEARCH

Figure 3. Stability of the immobilized receptor. (A) If either 11-cis- or 9-cis-retinal is supplemented in excess (10 µM), the activity staysconstant both for long time periods and during several activationcycles. The trace shows rhodopsin activity after complete photolysis(20 consecutive flashes) and regeneration with 9-cis-retinal. (B) If nochromophore is available, the activity of the receptor decaysexponentially with each flash, as the same fraction of the receptor isalways activated (under the chosen conditions, ∼ 30%). (C) Using asystem as in (A), the concentration of activated rhodopsin can becontrolled by the intensity of the flash. Experimental conditions: λ ≥ 495nm; transducin concentration = 2.5 µM; GTP concentration = 1 mM.

Figure 4. Ligand binding by the immobilized receptor. The receptorwas completely photolysed to opsin by 20 consecutive flashes. Then,11-cis-retinal was added to the indicated concentrations and allowedto bind for 40 min. Finally, the amount of generated rhodopsin wasdetermined by a flash (50% intensity). This was repeated for eachconcentration. (A) Recorded SPR traces, and (B) liganddose–response curve, with numbers referring to the correspondingSPR trace. The line corresponds to a Langmuir binding isotherm witha dissociation constant of 130 nM. Conditions were as in Figure 3.

A

B C

A B

© 1999 Nature America Inc. • http://biotech.nature.com©

199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://b

iote

ch.n

atu

re.c

om

mercaptododecanoic-(8-biotinoylamido-3,6-dioxaoctyl) amide (biotin-thiol) by following published procedures11. Streptavidin was fromCalbiochem-Novabiochem (San Diego, CA), GTP from Roche Diagnostics(Mannheim, Germany), biotin-aminocaprionic acid dihydrazide (BACH)and biotin hydrazide were from Fluka Chemical Corp. (Ronkonkoma, NY),9-cis-retinal was from Sigma (St. Louis, MO). Egg phosphatidylcholine(Fluka) was stored in chloroform at -20°C, dried under nitrogen, and redis-solved by ultrasonification in ROG50 just before use. Polydimethylsiloxanestamps for microcontact printing were kindly offered by Dr. E. Delamarche(IBM, Zürich, Switzerland), and 11-cis-retinal was a gift from Rosalie Crouch(Medical University of South Carolina, Charleston, and the National EyeInstitute, National Institute of Health, Bethesda, MD).

Preparation of washed membranes, transducin, and antirhodopsin anti-body. Rod outer segments were isolated from dark-adapted bovine retinas21,and washed membranes (rhodopsin in the native disk membranes but devoidof peripheral and soluble proteins) were prepared as described22. The mem-brane suspension was stored in 20 mM Bis-Tris Propane (1,3-bis[tris(hydroxymethyl)methylamino]propane), pH 7.5, 130 mM NaCl, and1 mM MgCl2 at -80°C until use. Transducin was prepared and stored asdescribed23. Monoclonal antibody 1D4 (ref. 24) directed against the C-termi-nal segment of rhodopsin was kindly prepared by the National Center forResearch Resources and the National Culture Center (Minneapolis, MN).

Self assembly and microcontact printing. For homogeneous surfaces, thesensor chips were immersed in a solution of 5 mM HTA/0.5 mM biotin-thiolin EtOH for >1 h immediately after gold evaporation. The ratio of 1:10between biotin-thiol and HTA has been shown to give the highest density ofimmobilized streptavidin11. As microcontact printing is driven by transportprocesses that are strongly influenced by the properties of the thiol25, usinginks containing mixtures of thiols (e.g., biotin-thiol/HTA) probably resultsin nonhomogeneous distribution of the different species in the printedregions. Therefore, we first printed the patterns of homogeneous HTA SAMsand then immersed the sample in a mixed biotin-thiol/HTA solution. Adroplet (∼ 200 µl) of 20 mM HTA in EtOH was put on the relief of the poly-dimethoxysilane stamp for 30 s and then blown away rapidly with a nitrogenstream. The stamp was deposited on the gold surface for 5 min withoutpressing. Then the sample was immersed for 10 s in 0.2 mM biotin-thiol/6mM HTA in EtOH, rinsed five times with 10 mM HTA in EtOH, and finallyimmersed for 2 min in the latter solution. This last step should displacebiotin-thiols that may have bound in the HTA regions to defective surfacestructures, where fast adsorption/desorption of thiols occurred.

Biotinylation of washed membranes. We followed a protocol for carbohy-drate-specific biotinylation of antibodies9. A 50 µM solution of rhodopsin inwashed membranes was incubated with 20 mM NaIO4 and 1.5 mg/ml (4 mM)BACH or 2 mg/ml (7.5 mM) biotin-hydrazide for 1 h at room temperature(reaction volume, 150 µl). Then the sample was dialyzed (Pierce SlidealizerCassette, 10 kDa cutoff; Pierce Chemical Co., Rockford, IL) at 4°C first for 5h,exchanging the buffer after each hour, and finally for 12h against 500 ml bufferR. Just before addition into the cuvette, the sample was diluted 1:5 in bufferROG50. Both biotinylation reagents showed indistinguishable performancesfor immobilization and ability of the receptor to activate transducin.

Experimental procedures. Measurements were performed in a stirredcuvette (250 µl volume). Instrumentation and analysis were as described7,using a xenon arc lamp (150 W) with an interference filter (750 ± 12 nm) formeasurements. Activation of rhodopsin was achieved by a photoflash usingeither a Schott (Wayzata, MN) OG530 or OG495 filter (λ ≥ 530 nm and λ ≥495 nm, respectively) to avoid photoreconversion of activated rhodopsin tothe ground state26. To reduce its intensity, gray filters (Schott) were fixed ontop of the wavelength filters.

AcknowledgmentsWe thank Dr. A. Brecht, D. Stamou, and Dr. R. Hovius for advice and criticalreading of the manuscript; A. Heusler for the synthesis of the alkane thiols; andDr. E. Delamarche (IBM, Zürich, Switzerland) for offering microcontact print-ing stamps. This research was supported by the Board of the Swiss FederalInstitutes of Technology (SPP MINAST, 7.06) to H.V., the DeutscheForschungsgemeinschaft (SFB 449) to K.P.H. and O.P.E., and the Fonds derChemischen Industrie to K.P.H.

1. Gudermann, T., Kalkbrenner, F. & Schultz, G. Diversity and selectivity of receptor-G protein interaction. Annu. Rev. Pharmacol. Toxicol. 36, 429–459 (1996).

2. Stadel, J.M., Wilson, S. & Bergsma, D.J. Orphan G protein-coupled receptors: aneglected opportunity for pioneer drug discovery. Trends Pharmacol. Sci. 18,430–437 (1997).

3. Helmreich, E.J.M. & Hofmann, K.-P. Structure and function of proteins in G pro-tein-coupled signal transfer. Biochim. Biophys. Acta. 1286, 285–322 (1996).

4. Fivash, M., Towler, E.M. & Fisher, R.J. BIAcore for macromolecular interaction.Curr. Opin. Biotechnol. 9, 97–101 (1998).

5. Hermanson, G.T., Mallia, A.K. & Smith, P.K. Immobilized affinity ligand tech-niques. (Academic Press, San Diego, CA; 1992).

6. Bruckert, F., Vuong, T.M. & Chabre, M. Light and GTP dependence of transducinsolubility in retinal rods. Eur. Biophys. J. 16, 207–218 (1988).

7. Heyse, S., Ernst, O.P., Dienes, Z., Hofmann, K.-P. & Vogel, H. Incorporation ofrhodopsin in laterally structured supported membranes: observation of trans-ducin activation with spatially and time-resolved surface plasmon resonance.Biochemistry 37, 507–522 (1998).

8. Vanrhee, A.M. & Jacobson, K.A. Molecular architecture of G-Protein-coupledreceptors. Drug Dev. Res. 37, 1–38 (1996).

9. O’Shannessy, D.J. & Quarles, R.H. Specific conjugation reactions of the oligosac-charide moieties of immunoglobulins. J. Appl. Biochem. 7, 347–355 (1985).

10. Bain, C.D. & Whitesides, G.M. Modeling organic surfaces with self-assembledmonolayers. Angew. Chem. Int. Ed. 101, 522–528 (1989).

11. Spinke, J. et al. Molecular recognition at self-assembled monolayers - optimiza-tion of surface functionalization. J. Chem. Phys. 99, 7012–7019 (1993).

12. Xia, Y. & Whitesides, G.M. Soft lithography. Angew. Chem. Int. Ed. 37, 550–575(1998).

13. Sackmann, E. Supported membranes: scientific and practical applications.Science 271, 43–48 (1996).

14. Lang, H., Duschl, C. & Vogel, H. A new class of thiolipids for the attachment oflipid bilayers on gold surfaces. Langmuir 10, 197–210 (1994).

15. Farrens, D.L. & Khorana, H.G. Structure and function in rhodopsin. Measurementof the rate of metarhodopsin II decay by fluorescence spectroscopy. J. Biol.Chem. 270, 5073–5076 (1995).

16. Antonny, B., Otto-Bruc, A., Chabre, M. & Vuong, T.M. GTP hydrolysis by purifiedalpha-subunit of transducin and its complex with the cyclic GMP phosphodi-esterase inhibitor. Biochemistry 32, 8646–8653 (1993).

17. Freissmuth, M., Waldhoer, M., Bofill-Cardona, E. & Nanoff, C. G protein antago-nists. Trends Pharmacol. Sci. 20, 237–245 (1999).

18. Hamm, H.E. et al. Site of G protein binding to rhodopsin mapped with syntheticpeptides from the alpha subunit. Science 241, 832–835 (1988).

19. Hovius, R. et al. Fluorescence techniques for fundamental and applied studies ofmembrane protein receptors: the 5-HT3 serotonin receptor. J. Recept. SignalTransduct. Res. 19, 533–545 (1999).

20. Remmers, A.E., Posner, R. & Neubig, R.R. Fluorescent guanine nucleotideanalogs and G protein activation. J. Biol. Chem. 269, 13771–13778 (1994).

21. Papermaster, D.S. Preparation of antibodies to rhodopsin and the large protein ofrod outer segments. Methods Enzymol. 81, 240–246 (1982).

22. Kühn, H. Interaction of the rod cell proteins with the disk membrane: influence oflight, ionic strength and nucleotides. Curr. Top. Membr. Transp. 15, 171–201(1981).

23. Heck, M. & Hofmann, K.P. G-protein-effector coupling: a real-time light-scatter-ing assay for transducin-phosphodiesterase interaction. Biochemistry 32,8220–8227 (1993).

24. Molday, R.S. & MacKenzie, D. Monoclonal antibodies to rhodopsin: characteriza-tion, cross-reactivity, and application as structural probes. Biochemistry 22,653–660 (1983).

25. Delamarche, E. et al. Transport mechanisms of alkanethiols during microcontactprinting on gold. J. Phys. Chem. B 102, 3324–3334 (1998).

26. Arnis, S. & Hofmann, K.P. Photoregeneration of bovine rhodopsin from its signal-ing state. Biochemistry 34, 9333–9340 (1995).

1108 NATURE BIOTECHNOLOGY VOL 17 NOVEMBER 1999 http://biotech.nature.com

RESEARCH

© 1999 Nature America Inc. • http://biotech.nature.com©

199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://b

iote

ch.n

atu

re.c

om