ANALYTICALBIOCHEMISTRY

Analytical Biochemistry 355 (2006) 132–139

www.elsevier.com/locate/yabio

Analyzing ligand and small molecule binding activity of solubilized GPCRs using biosensor technology

Iva Navratilova a, Marianna Dioszegi b, David G. Myszka a,¤

a Center for Biomolecular Interaction Analysis, University of Utah School of Medicine, Salt Lake City, UT 84132, USAb Roche Palo Alto LLC, Palo Alto, CA 94304, USA

Received 17 February 2006Available online 15 May 2006

Abstract

We used Biacore technology to measure directly the binding of natural ligands and small molecules to the chemokine receptorsCXCR4 and CCR5. Both G protein-coupled receptors were solubilized from whole cell pellets and captured on antibody surfaces foranalysis. Our solubilization conditions maintained high-aYnity binding of chemokines SDF-1� and RANTES to CXCR4 and CCR5,respectively. Surface density- and buVer-dependent binding responses, along with binding data for a selective ligand (RCP-168), furthervalidated the biosensor assay. In addition, we showed that it is possible to collect high-quality binding responses for the archetypal smallmolecule inhibitors JM-2987 and TAK-779. Finally, using our biosensor-based method, we characterized the kinetics of 19 novel smallmolecule inhibitors of CCR5 and showed that their aYnities correlated with values determined for the membrane-associated receptor.Together, the chemokine and small molecule binding data provide evidence that the solubilized receptors maintain native binding proper-ties. These solubilized receptor preparations could be useful reagents for biophysical studies as well as for structural analysis.© 2006 Elsevier Inc. All rights reserved.

Keywords: Biacore; Surface plasmon resonance; Protein–protein interaction; Kinetics; SPR

Surface plasmon resonance (SPR)1 biosensors havebecome standard tools for characterizing protein interactions[1–3]. One of the emerging application areas is the use of bio-sensor technology to study membrane-associated receptors.In one approach, Karlsson and Löfås immobilized a puriWed

* Corresponding author. Fax: +1 801 585 3015.E-mail address: dmyszka@cores.utah.edu (D.G. Myszka).

1 Abbreviations used: SPR, surface plasmon resonance; GPCR, G pro-tein-coupled receptor; DOPC, 1,2-dioleoyl-sn-glycero-3-phosphocholine;DOPS, 1,2-dioleoyl-sn-glycero-3-phospho-L-serine (sodium salt); DOM,n-dodecyl-�-D-maltopyranoside; Chaps, 3-[(3-cholamidopropyl)-dimeth-ylammonio]-1-propane sulfonate/N,N-dimethyl-3-sulfo-N-[3-[[(3�,5�,7�,12�)-3,7,12-trihydroxy-24- oxocholan-24-yl]amino]propyl]-1-propanaminiumhydroxide (inner salt); CHS, cholesteryl hemisuccinate tris salt; Hepes, N-(2-hydroxyethyl)piperazine-N�-(2-ethanesulfonic acid); HBS-P, Hepes-buVeredsaline with P20; EDC, 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hy-drochloride; NHS, N-hydroxy succinimide; BSA, bovine serum albumin;SMM, synthetically and modularly modiWed; DMSO, dimethyl sulfox-ide; FBS, fetal bovine serum; FACS, Xuorescence-activated cell sorting;PBS, phosphate-buVered saline; PEI, polyethyleneimine.

0003-2697/$ - see front matter © 2006 Elsevier Inc. All rights reserved.doi:10.1016/j.ab.2006.04.021

receptor onto the sensor surface and then reconstituted amembrane environment on this surface and conWrmed that itwas active for binding [4]. We extended this method by show-ing that it was possible to capture receptors out of crudepreparations and study directly the binding of antibodies [5].More recently, we demonstrated that it was possible to moni-tor the binding activity of solubilized receptors without lipidreconstitution. We found that this approach is an idealmethod for identifying solubilization conditions that main-tain receptor binding activity [6].

Our initial receptor work focused on high-molecularweight ligands such as monoclonal antibodies. By improv-ing the receptor solubilization conditions, we were able tocharacterize the binding of proteins such as HIV gp120 toits coreceptors (CXCR4 and CCR5). In addition, weshowed that the binding of gp120 was CD4 dependent, asexpected [7].

In this current article, we demonstrate that it is possibleto monitor the binding of native ligands and low-molecular

Analyzing ligand and small molecule binding activity / I. Navratilova et al. / Anal. Biochem. 355 (2006) 132–139 133

weight compounds to our solubilized G protein-coupledreceptors (GPCRs). Fig. 1 shows the basic principle of ourapproach. Receptors expressed with a C-terminal peptidetag (C9) [8] are captured onto an antibody (1D4) surface[5]. Capturing allows us to immobilize receptors from crudepreparations without prior puriWcation steps. Receptorbinding activity is then tested using conformationallydependent antibodies and proteins (gp120/CD4) [6] as wellas natural chemokine ligands and low-molecular-weightinhibitors.

The biosensor assay is rapid and can generate high-reso-lution information about a wide range of binding interac-tions. By comparing the aYnities determined for thesolubilized receptor with those determined from mem-brane-associated receptor assays, we are able to demon-strate that our solubilized receptor preparations retainnative binding properties. These reagents may provideexcellent starting material for structural analysis and thebiosensor methods should be readily adaptable to otherreceptor systems.

Materials and methods

Materials

Biacore 2000 and S51 optical biosensors, CM4 sensorchips, and the amine-coupling kit were obtained from Bia-core AB (Uppsala, Sweden). 1D4 antibody was purchasedfrom the University of British Columbia. The small mole-cule inhibitors TAK-779 and JM-2987 were obtained fromthe AIDS Research and Reference Reagent Program (Divi-sion of AIDS, National Institutes of Health), and the other

Fig. 1. Design of the biosensor assay. 1D4-capturing antibody is immobi-lized on the carboxydextran matrix that coats the sensor chip surface anddetergent-solubilized GPCR expressed with a C-terminal tag is capturedby the 1D4 antibody. Activity binding tests are performed using confor-mation-dependent antibodies and CD4/gp120 protein complexes. Cap-tured receptor is also tested for natural chemokine and small moleculeinhibitor binding.

small molecule inhibitors were obtained from Roche (PaloAlto, CA, USA). The recombinant human chemokinesSDF-1� and RANTES were obtained from R&D Systems(Minneapolis, MN, USA), and the SMM chemokine RCP-168 was purchased from Raylight (La Jolla, CA, USA). Thehuman chemokine receptors CXCR4 and CCR5 were over-expressed in Cf2Th canine thymocyte cells as described pre-viously [9]; the cells were propagated by the National CellCulture Center, and both receptors contained a C-terminallinear C9 peptide tag (TETSQVAPA) that is recognized bythe 1D4 monoclonal antibody [8]. Lipids (syntheticphospholipid blend [Dioleoyl] DOPC:DOPS [7:3, w/w]), aMini-Extruder kit, and polycarbonate Wlters (100 nm) werepurchased from Avanti Polar Lipids (Alabaster, AL, USA).Detergents (n-dodecyl-�-D-maltopyranoside [DOM],n-octyl-�-D-glucopyranoside, and Chaps) were purchasedfrom Anatrace (Maumee, OH, USA). Cholesteryl hemisuc-cinate tris salt (CHS) was purchased from Sigma–Aldrich(St. Louis, MO, USA), and complete, EDTA-free, proteaseinhibitor tablets (cat. no. 11-697-498-001) were purchasedfrom Roche Diagnostics (Indianapolis, IN, USA).

Preparation of lipids

To prepare lipid/detergent mixed micelles, aliquots ofseveral lipids (5 mM Wnal concentration) were transferredinto glass test tubes. A thin lipid Wlm was formed on theglass wall by rotating the tube while evaporating the chlo-roform using a stream of nitrogen gas. Any remaining chlo-roform was removed under vacuum for at least 1 h. Then1 ml of buVer (50 mM Hepes, 150 mM NaCl, pH 7.0) wasadded to the dry lipid Wlms. The mixture was frozen,thawed, and vortexed four times. Uniform liposomes wereprepared by extrusion through a polycarbonate Wlter ofdeWned pore diameter (100 nm) using an Avanti Mini-Extruder kit. Detergent/lipid micelles, which were used forsolubilization trials, were prepared by adding the 2% CHS/10% DOM/10% Chaps in a 1:2 detergent/lipid ratio andvortexing the sample.

Immobilization of 1D4 monoclonal antibody on a CM4 surface

The monoclonal antibody 1D4 was immobilized on aCM4 sensor chip using standard amine-coupling chemistry.HBS-P (10 mM Hepes, 0.15 M NaCl, 0.005% P20, pH 7.4)was used as the running buVer. The carboxymethyl dextransurface was activated with a 7-min injection of a 1:1 ratio of0.4 M 1-ethyl-3-(3-dimethylaminopropyl) carbodiimidehydrochloride (EDC)/0.1 M N-hydroxy succinimide (NHS).The antibody was coupled to the surface with a 7-min injec-tion of 1D4 diluted in 10 mM sodium acetate (pH 5.5).Remaining activated groups were blocked with a 7-mininjection of 1 M ethanolamine (pH 8.5). To obtain a 1D4surface density of approximately 7000 RU, immobilizationwas performed at 25 °C using Biacore 2000 and at 32 °Cusing Biacore S51.

134 Analyzing ligand and small molecule binding activity / I. Navratilova et al. / Anal. Biochem. 355 (2006) 132–139

Solubilization of receptor

Cf2Th–CXCR4 and Cf2Th–CCR5 cells were solubilizedin a buVer of 20 mM Tris (pH 7.0), 0.1 M (NH4)2SO4, 10%glycerol, 0.07% CHS, 0.33% DOM, 0.33% Chaps, 0.33 mMDOPC/DOPS (7:3), and 1 protease inhibitor tablet per50 ml buVer. For solubilization trials, 0.5 ml of this solubili-zation buVer was added to approximately 2£ 106 cells, andthese cell suspensions were sonicated using a probe sonica-tor (6£ 1-s pulses) and placed on a rocker at 4 °C. After 2 h,the solutions were centrifuged at 4 °C for 20 min at14,000 rpm using a tabletop centrifuge. The supernatantscontaining CXCR4 and CCR5 were transferred to newtubes and kept frozen at ¡80 °C until analysis.

Ligand binding to receptors

Solubilized CCR5 and CXCR4 were captured by 1D4immobilized on surfaces within a CM4 chip. ChemokineSDF-1� and RANTES were injected over the capturedreceptor at concentrations of 33, 11, 3.67, 1.22, 0.47, and0.14 nM at a Xow rate of 90 or 100 �l/min. Association wasmonitored for 1 min, and dissociation was monitored for10 min. Each analyte injection was followed by a blankinjection to reference for drift caused by the decaying recep-tor surface. RANTES was diluted in running buVer con-taining 50 mM Hepes (pH 7.0), 0.02% CHS, 0.1% DOM,0.1% Chaps, and 50 nM DOPC/DOPS (7:3). For the SDF-1�/CXCR4 studies, 5 mM MgCl2, 1 mM CaCl2, and 0.1 mg/ml bovine serum albumin (BSA) were added to this buVer.SDF-1� binding was tested at three diVerent concentrationsof NaCl (50, 75, and 150 mM). RANTES binding was mea-sured in 150 mM NaCl over CCR5 surfaces prepared atthree densities: 1000, 2000, and 4000 RU. The nonnaturalsynthetically and modularly modiWed (SMM) chemokineRCP-168 was diluted in a running buVer containing150 mM NaCl and was tested in a concentration series of9.0, 3.0, 1.0, 0.33, and 0.11 nM over CXCR4 and CCR5 sur-faces built on one chip. Association was monitored for3 min, and dissociation was monitored for 10 min. Theinjections were performed from the lowest to the highestconcentration so as to minimize accumulation of analyteon the reaction surface. No regeneration was performedbetween analyte injections. Experiments were carried out at20–25 °C.

Binding of small molecule inhibitors to receptors

CCR5 and CXCR4 were captured at densities ofapproximately 3000 RU on a surface of a 1D4-coated CM4chip. For water-soluble compounds (TAK-779 and JM-2987), the running buVer consisted of 50 mM Hepes (pH7.0), 150 mM NaCl, 0.02% CHS, 0.1% DOM, 0.1% Chaps,and 50 nM DOPC/DOPS (7:3); for the CXCR4 studies,5 mM MgCl2 and 1 mM CaCl2 were added to this buVer.The 19 small molecule CCR5 compounds (averageMWD550 Da) were dissolved in dimethyl sulfoxide

(DMSO) and diluted into 50 mM Hepes (pH 7.0), 150 mMNaCl, 0.02% CHS, 0.1% DOM, 0.1% Chaps, and 50 nMDOPC/DOPS (7:3) to concentrations ranging from 33 �M(for the weakest-aYnity inhibitor) to 1.7 nM (for the high-est-aYnity inhibitor). These solutions were then seriallydiluted threefold in running buVer (50 mM Hepes [pH 7.0],150 mM NaCl, 0.02% CHS, 0.1% DOM, 0.1% Chaps, 50 nMDOPC/DOPS [7:3], 0.5% DMSO) to produce the concen-tration series for each inhibitor. Given the higher sensitivityof the Biacore S51, this instrument was used to assay the 19novel inhibitors of CCR5. The receptor surfaces were stabi-lized by at least 10 start-up buVer blanks before injecting acompound. To reference for drift, two blank injections wereperformed between analyte injections. Compounds wereinjected at a Xow rate of 30 �l/min, and the association anddissociation phases were monitored for 1–3 and 3–10 minfor the low- and high-aYnity compounds, respectively. Thereceptor surfaces were not regenerated between analyteinjections. Instead, the injections were performed from low-est to highest concentrations and the data were normalizedfor the maximum binding capacity.

hCCR5 expression plasmid

The hCCR5 receptor ORF was subcloned using PCRfrom OriGene’s TrueClone (cat. no. TC110858, OriGeneTechnologies, Rockville, MD, USA). The following PCRprimers were used in reaction: 5� primer (5�-ATA-TAT-TAA-TCT-AGA-ACC-ATG-GAT-TAT-CAA-GTGTCA-AGT-C-3�) and 3� primer (5�-ATA-TAT-TCT-AGA-GCG-GAT-CCT-CAC-AAG-CCC-ACA-GAT-ATT-TC-3�).PCR was carried out with the following cycling parameters(94 °C for 1 min, 55 °C for 1 min, and 72 °C for 2 min) for 30cycles and with Roche Expand High Fidelity PCR System(Roche Applied Science, Indianapolis, IN, USA). The 1.1-kb PCR product was digested with XbaI and BamHI andwas cloned into the mammalian expression vectorpcDNA3.1(–) (Invitrogen, Carlsbad, CA, USA). The cloneswere sequence conWrmed and found to contain the identicalhCCR5 coding sequence (GenBank Accession No.NM_000579).

Stable expression of wild-type CCR5 in CHO–G�16 cells

CHO–G�16 cells were plated in F12 medium supple-mented with 10% fetal bovine serum (FBS) and 200 �g/mlhygromycin overnight at 37 °C in a CO2 incubator prior totransfection. FuGene 6 (Roche Applied Science) transfec-tion reagent was used to transfect the expression plasmidper the manufacturer’s instructions. In brief, plasmidDNAs were mixed with FuGene 6 reagent diluted withserum-free medium. The mixture was incubated at roomtemperature for 15 min. After incubation, the mixture wasadded to the cells at approximately 60% conXuency. Next,48 h after transfection, a selection marker G418 wasintroduced into the medium to the Wnal concentration of1 mg/ml. The population of cells stably expressing CCR5

Analyzing ligand and small molecule binding activity / I. Navratilova et al. / Anal. Biochem. 355 (2006) 132–139 135

was enriched by several rounds of Xuorescence-activatedcell sorting (FACS) using 2D7 monoclonal hCCR5 anti-body (BD Bioscience Pharmigen, San Diego, CA, USA).

Radioligand binding assays

125I RANTES with a speciWc activity of approximately2200 Ci/mmol was purchased from Perkin-Elmer Life Sci-ences (Boston, MA, USA). Whole cell binding assay onCHO–G�16–hCCR5 wild-type cells was performed as fol-lows. Adherent CHO–G�16–hCCR5 cells stably expressinghCCR5 receptor and G protein G�16 were grown toapproximately 90% conXuency. The cells were detached infreshly made phosphate-buVered saline (PBS) containing1 mM EDTA. Cells were washed twice in PBS without Ca2+

and Mg2+ and were resuspended in ice-cold binding buVer(phenol red-free F12 medium, pH 7.24, supplemented withfreshly made 0.1% BSA and 0.1% NaN3). Cells were platedin 96-well culture plates at 1.5£ 105 cells/well. Seriallydiluted, 10-fold dilutions (range 10¡4–10¡13 M) of CCR5inhibitors were added to the cells, followed by the additionof 100 pM of 125I RANTES. After 2 h of incubation atroom temperature with gentle shaking, cells were harvestedonto GF/C UniFilter plates (Perkin-Elmer Life Sciences)using a cell harvester. UniFilter plates were pretreated with0.3% polyethyleneimine (PEI)/0.2% BSA for 30 min prior toharvest. Filter plates were washed Wve times with cold25 mM Hepes buVer (pH 7.1) containing 500 mM NaCl,1 mM CaCl2, and 5 mM MgCl2. Plates were dried in a 70 °Coven for 20 min, 40 �l scintillation Xuid was added, andradioactivity was measured using a TopCount NXT (Per-kin-Elmer, Shelton, CT, USA). In all experiments, eachdata point was assayed in duplicate. Curve Wtting and sub-sequent data analysis were carried out using GraphPadPRISM software (Intuitive Software for Science, SanDiego, CA, USA), and EC50 values were calculated usingnonlinear regression analysis. PRISM calculates the KIusing the Cheng and PrusoV equation [10].

Results

Natural ligand binding to solubilized GPCRs

Cells expressing either CCR5 or CXCR4 were solubi-lized in a previously optimized solubilization buVer [6].Each crude preparation was then injected over a 1D4 anti-body surface for receptor capturing and binding tests asdescribed previously [6]. The responses in Fig. 2A showthe binding of the chemokine RANTES (MW »7800 Da)to CCR5 that was captured on the sensor chip surface atthree diVerent densities (4000, 2000, and 1000 RU). Theseresults illustrate that the responses observed for RANTES(e.g., 50, 25, and 10 RU for the highest RANTES concen-tration) correlated with the density of CCR5 captured oneach surfaces. We globally Wt the binding kinetics fromthe three diVerent density surfaces to a 1:1 interactionmodel that included a step for mass transport [11]. Under

the buVer conditions used in this assay, we found that theassociation and dissociation rates were 1.2£ 107 M¡1 s¡1

and 0.04 s¡1, respectively. These rate constants yield anaYnity for the RANTES/CCR5 interaction of 3.9 nM,which is consistent with aYnities reported in the literature(0.2–6.2 nM) for membrane-associated receptor assays[12,13].

Fig. 2B shows the binding responses for SDF-1� (MW»7800 Da) interacting with solubilized CXCR4 capturedon a sensor chip surface. In these experiments, SDF-1� wasprepared in buVers containing diVerent concentrations ofNaCl (75, 150, and 300 mM). For a better visual compari-son, responses were normalized for Rmax, which representsthe maximum binding capacity of the surface (determinedduring data analysis). It is apparent from a visual inspec-tion of the data that the aYnity of the SDF-1�/CXCR4interaction decreases with increasing salt concentrations.The responses collected within each buVer condition wereglobally Wt to a 1:1 interaction model that included a masstransport step. The apparent association and dissociationrate constants ranged from 4£ 108 to 2.9£ 106 M¡1 s¡1 and0.5 to 0.08 s¡1, respectively, and the NaCl concentrationranged from 75 to 300 mM, corresponding to aYnities of1.2 to 27 nM. SDF-1� is a highly basic ligand that report-edly binds to CXCR4 with aYnities in the range of 1–5 nMdepending on assay conditions [14,15]. The fast associationrate that we observed for the SDF-1�/CXCR4 interaction,as well as the strong salt dependence and good correlationin aYnity with membrane-associated receptor assays, pro-vides evidence that the solubilized receptor maintains natu-ral ligand binding properties. Importantly, we found thatwe could freeze/thaw solubilized preparations of bothCCR5 and CXCR4 without any detectable loss in ligandbinding activity (data not shown).

Selective binding of SMM chemokine RCP-168 to CXCR4 versus CCR5

RCP-168 is an SMM chemokine (MW »8100 Da) inwhich 10 amino acids at the N-terminus of viral macro-phage inXammatory protein II are replaced with the non-natural D-amino acids (D-[1-10]-vMIP-II). This chemokinereportedly binds with approximately 10-fold higher aYnityto CXCR4 than to CCR5 [16,17]. Fig. 2C shows theresponse data collected for RCP-168 injected simulta-neously over CXCR4 and CCR5 surfaces. We observedconcentration-dependent binding responses over theCXCR4 surface, whereas the binding levels on the CCR5surface were signiWcantly lower. The response data were Wtto a 1:1 interaction model that included a mass transportstep to yield an aYnity of 0.4 nM for RCP-168 interactingwith the CXCR4 surface and approximately 7 nM for theCCR5 surface. The diVerence in binding aYnity of RCP-168 for the two solubilized receptors is consistent with pre-vious results [16,17] and further demonstrates that thereceptor preparations maintain their ability to bind ligandsselectively.

136 Analyzing ligand and small molecule binding activity / I. Navratilova et al. / Anal. Biochem. 355 (2006) 132–139

Binding of small molecule inhibitors JM-2987 and TAK-779 to solubilized GPCRs

One of the challenges in the application of biosensortechnology to study membrane-associated receptors is todetect directly the binding of small molecule ligands.Because the signal associated with the small molecule willbe small due to the SPR signal being proportional to mass,

we need to use receptor preparations that maintain highlevels of activity. The data represented in Fig. 2D show thebinding responses of the small molecule inhibitors JM-2987and TAK-779 to CXCR4 and CCR5, respectively. Bothdata sets Wt well to a 1:1 interaction model and producedbinding constants of 2 and 380 nM for JM-2987 and TAK-779, respectively. The aYnity obtained for the JM-2987/CXCR4 interaction is consistent with values reported in the

Fig. 2. (A) Responses for RANTES (33, 11, 3.7, 1.2, and 0.4 nM) binding to CCR5 captured on 1D4 antibody surface at three densities: 4000, 2000, and1000 RU (note that the plots share the same y axis). (B) Responses for SDF-1� (33, 11, 3.7, 1.2, and 0.4 nM) binding to captured CXCR4 in the presence ofthree diVerent NaCl concentrations: 75, 150, and 300 mM (note that the plots share the same y axis). (C) Binding responses of SMM chemokine RCP-168(9, 3, 1, 0.3, and 0.1 nM) to CXCR4 and CCR5 captured on diVerent surfaces of one CM4 chip. The inset shows injections of RCP-168 at concentrations of247, 82, 27, 9, 3, 1, 0.3, and 0.1 nM. (D) Responses of small molecule inhibitors JM-2987 (136, 45, 15, 5, and 1.7 nM) and TAK-779 (30, 10, 3.3, 1.1, 0.4, and0.1 �M in duplicate injections) binding to CXCR4 and CCR5 receptors. All data were collected from Biacore 2000 and Wt to a 1:1 interaction model. Fitsare represented by red curves. (For interpretation of the references to color in this Wgure legend, the reader is referred to the Web version of this article.)

0 80 160

0

7

14

Res

pons

e (R

U)

0 200 400

0

60

120

Nor

mal

ized

Res

pons

e

Time (s)

D

A

0 100 200

50

100

0

100 2000 2000 100

Nor

mal

ized

Res

pons

e

B

75 mM NaCl 150 mM NaCl 300 mM NaCl

SDF-1α/CXCR4

0 100 200

25

50

0

0 100 200 0 100 200

Res

pons

e (R

U)

4000 RU 2000 RU 1000 RU

RANTES/CCR5

Small molecules

JM-2987/CXCR4 TAK-779/CCR5

C

0 200 400

0

50

100

Nor

mal

ized

Res

pons

e

SMM chemokine RCP-168

CXCR4 CCR5

0 200 400

0 200 400

0

20

40

Analyzing ligand and small molecule binding activity / I. Navratilova et al. / Anal. Biochem. 355 (2006) 132–139 137

literature (»5 nM) [18]. The aYnity for the TAK-779/CCR5 interaction, however, is weaker than expected basedon literature values (3–28 nM) [19,20]. To determinewhether this diVerence in aYnity was systemic, we charac-terized the binding of an additional 19 small molecules toCCR5 for which we had cell binding data available to com-pare with the biosensor results.

Binding of 19 small molecule inhibitors to CCR5

Using the Biacore assay, we studied the binding of 19novel small molecule inhibitors (average MW » 550 Da) tosolubilized CCR5 (Fig. 3). The compounds were injectedover freshly prepared CCR5 surfaces in increasing concen-trations to eliminate regeneration steps. Because there is noregeneration during the compound injection series, thebound analyte can accumulate on the receptor, therebydecreasing its binding capacity. Therefore, all bindingresponses were globally Wtted with a 1:1 interaction modelthat used a diVerent maximum binding capacity (Rmax) foreach analyte injection and then data were normalized forRmax. This normalization increases the apparent noise levelof the sensorgram for the highest concentrations of thehigh-aYnity compounds. To visualize diVerences in bindingbehavior between compounds, the sets of sensorgrams inFig. 3 are arranged in order of decreasing aYnity.

Correlating the aYnity determined for solubilized and membrane-associated receptor

These 19 compounds were also analyzed for binding tomembrane-associated CCR5 using a whole-cell radioliganddisplacement assay. In these experiments, the ability of eachcompound to block binding of radiolabeled RANTES wasused to determine KI. The data shown in Fig. 4 plot the KIobserved from the whole cell assay against the KD deter-

mined for the solubilized receptor. We found that the corre-lation in aYnities determined from the two assays wasremarkably good. The excellent trend over a 10,000-foldrange in aYnity from weak to strong binders provides evi-dence that our solubilization conditions maintain theintrinsic binding properties of the receptor.

Discussion

We showed previously that it was possible to solubilizeand capture receptors on the sensor chip surface and tomeasure the binding of relatively large biomolecules, suchas antibodies and gp120/CD4 complex, without the need toreconstitute a lipid bilayer [6]. We also demonstrated previ-ously how the biosensor can be used as a screening tool to

Fig. 4. Correlation plot of aYnity constants (KD values) measured usingBiacore biosensors versus inhibition constants (KI values) obtained bywhole cell-based experiments for 19 small molecule inhibitors binding toCCR5. The correlation coeYcient is 0.785.

KD[M

]S

olub

ilize

dre

cept

or

KI[M]Membrane-associated receptor

10-5 10-6 10-7 10-8 10-9

10-5

10-6

10-7

10-8

10-9

8

9

1

4

10

6

5

16

7

32

12

19

1517

1113 14

18

Fig. 3. Responses of 19 small molecule inhibitors binding to CCR5. Compounds were injected in threefold dilutions. The highest concentrations injectedwere 1.22 �M for compounds 1, 2, and 7; 0.4 �M for compounds 3 and 8; 3.7 �M for compounds 4, 5, 6, and 13; 11 �M for compounds 9, 10, 11, 12, 14, and18; and 33 �M for compounds 15, 16, 17, and 19. The data sets are arranged from the highest to the lowest aYnity interactions. Data were collected usingBiacore S51.

1 2 3 4 5

6 7 8 9 10

11 12 13 14 15

16 17 18 19

138 Analyzing ligand and small molecule binding activity / I. Navratilova et al. / Anal. Biochem. 355 (2006) 132–139

identify suitable solubilization conditions and determinethe eVects of temperature, running buVer composition,detergents, pH, and the like on receptor activity [6]. Forexample, we found that the most eVective detergent forCCR5 and CXCR4 solubilization was a mixture of CHS,DOM, and Chaps. Also, the presence of the lipids DOPCand DOPS had a signiWcant inXuence on the stability of sol-ubilized receptors. Recently, we showed how the biosensorcould be used to help identify crystallization conditionsthat promote protein complex formation [21]. In this study,we showed that receptors can now be captured in a highlyactive conformation, making it possible to measure bindingof natural chemokines and small molecule inhibitors.

Our current results suggest that the presence of appro-priate lipids and detergents suYciently mimics the naturalenvironment of the membrane-associated receptors andtherefore maintains them in active conformations for anextended period of time. The binding aYnities of naturalchemokines SDF-1� and RANTES to CXCR4 and CCR5,respectively, correspond well to the data reported in litera-ture from whole cell studies. In addition, the receptor selec-tivity of the nonnatural SMM chemokine RCP-168 wasconWrmed for both CXCR4 and CCR5 receptors. We alsofound that we can freeze/thaw solubilized receptor prepara-tions, thereby, simplifying the preparation of samples formore extensive puriWcation procedures.

Once we optimized the activity of the solubilized recep-tor, we showed that it was possible to measure directly thebinding of low-molecular-weight compounds. We charac-terized the binding of 19 novel low-molecular-weight inhib-itors of CCR5 that were chosen because they spanned awide range of aYnities. We showed that the aYnities deter-mined for the solubilized receptor correlated well with inhi-bition constants determined in cell-based assays. Thisindicates that the solubilized receptor maintains ligandbinding properties similar to those of the membrane-associ-ated receptor.

In a myopic view, our work provides a new method forstudying the binding properties of natural ligands and smallmolecules to two important chemokine receptors. Steppingback a little, we see that having solubilized, active, and stableforms of these receptors permits more detailed biophysicalanalyses as well as the potential for structural characterization.Taking a panoramic view, we see how biosensor assays can beused to identify solubilization conditions for membrane-asso-ciated proteins as well as to characterize their activity. As bio-sensor technology evolves, we expect to see its application tomembrane receptor research expand considerably.

Acknowledgments

This work was supported by funding from the NationalInstitutes of Health (P01 GM66521 to DGM). We thankTony Giannetti and David M. Rotstein at Roche (PaloAlto, CA, USA) for providing 19 small molecule com-pounds for analysis. The following reagents were obtainedthrough the AIDS Research and Reference Reagent Pro-

gram (Division of AIDS, National Institutes of Health):TAK-779 and bicyclam JM-2987.

References

[1] R.L. Rich, D.G. Myszka, Survey of the year 2004 commercial opticalbiosensor literature, J. Mol. Recognit. 18 (2005) 431–478.

[2] R.L. Rich, D.G. Myszka, Survey of the year 2003 commercial opticalbiosensor literature, J. Mol. Recognit. 18 (2005) 1–39.

[3] R.L. Rich, D.G. Myszka, A survey of the year 2002 commercial opti-cal biosensor literature, J. Mol. Recognit. 16 (2003) 351–382.

[4] O.P. Karlsson, S. Löfås, Flow-mediated on-surface reconstitution ofG-protein coupled receptors for applications in surface plasmon reso-nance biosensors, Anal. Biochem. 300 (2002) 132–138.

[5] P. Stenlund, G.J. Babcock, J. Sodroski, D.G. Myszka, Capture andreconstitution of G protein-coupled receptors on a biosensor surface,Anal. Biochem. 316 (2003) 243–250.

[6] I. Navratilova, J. Sodroski, D.G. Myszka, Solubilization, stabilization,and puriWcation of chemokine receptors using biosensor technology,Anal. Biochem. 339 (2005) 271–281.

[7] H. Choe, K.A. Martin, M. Farzan, J. Sodroski, N.P. Gerard, C. Ger-ard, Structural interactions between chemokine receptors, gp120 Envand CD4, Semin. Immunol. 10 (1998) 249–257.

[8] D.D. Oprian, R.S. Molday, R.J. Kaufman, H.G. Khorana, Expressionof a synthetic bovine rhodopsin gene in monkey kidney cells, Proc.Natl. Acad. Sci. USA 84 (1987) 8874–8878.

[9] T. Mirzabekov, N. Bannert, M. Farzan, W. Hofmann, P. Kolchinsky,L. Wu, R. Wyatt, J. Sodroski, Enhanced expression, native puriWca-tion, and characterization of CCR5, a principal HIV-1 coreceptor, J.Biol. Chem. 274 (1999) 28745–28750.

[10] Y. Cheng, W.H. PrusoV, Relationship between the inhibition constant(K1) and the concentration of inhibitor which causes 50 per cent inhi-bition (I50) of an enzymatic reaction, Biochem. Pharmacol. 22 (1973)3099–3108.

[11] D.G. Myszka, X. He, M. Dembo, T.A. Morton, B. Goldstein, Extendingthe range of rate constants available from Biacore: interpreting masstransport-inXuenced binding data, Biophys. J. 75 (1998) 583–594.

[12] S. Polo, V. Nardese, C. De Santis, C. Arcelloni, R. Paroni, F. Sironi, A.Verani, M. Rizzi, M. Bolognesi, P. Lusso, Enhancement of the HIV-1inhibitory activity of RANTES by modiWcation of the N-terminalregion: dissociation from CCR5 activation, Eur. J. Immunol. 30(2000) 3190–3198.

[13] C. Blanpain, I. Migeotte, B. Lee, J. Vakili, B.J. Doranz, C. Govaerts,G. Vassart, R.W. Doms, M. Parmentier, CCR5 binds multiple CC-chemokines: MCP-3 acts as a natural antagonist, Blood 94 (1999)1899–1905.

[14] M.P. Crump, J.H. Gong, P. Loetscher, K. Rajarathnam, A. Amara,F. Arenzana-Seisdedos, J.L. Virelizier, M. Baggiolini, B.D. Sykes,I. Clark-Lewis, Solution structure and basis for functional activityof stromal cell-derived factor-1: dissociation of CXCR4 activa-tion from binding and inhibition of HIV-1, EMBO J. 16 (1997)6996–7007.

[15] J. Di Salvo, G.E. Koch, K.E. Johnson, A.D. Blake, B.L. Daugherty,J.A. DeMartino, A. Sirotina-Meisher, Y. Liu, M.S. Springer, M.A.Cascieri, K.A. Sullivan, The CXCR4 agonist ligand stromal derivedfactor-1 maintains high aYnity for receptors in both G�i-coupled anduncoupled states, Eur. J. Pharmacol. 409 (2000) 143–154.

[16] C. BoshoV, Y. Endo, P.D. Collins, Y. Takeuchi, J.D. Reeves, V.L.Schweickart, M.A. Siani, T. Sasaki, T.J. Williams, P.W. Gray,P.S. Moore, Y. Chang, R.A. Weiss, Angiogenic and HIV-inhibi-tory functions of KSHV-encoded chemokines, Science 278 (1997)290–294.

[17] M. Mori, D. Liu, S. Kumar, Z. Huang, NMR structures of anti-HIVD-peptides derived from the N-terminus of viral chemokine vMIP-II,Biochem. Biophys. Res. Commun. 335 (2005) 651–658.

[18] E. De Clercq, The bicyclam AMD3100 story, Nat. Rev. Drug Discov.2 (2003) 581–587.

Analyzing ligand and small molecule binding activity / I. Navratilova et al. / Anal. Biochem. 355 (2006) 132–139 139

[19] M. Baba, O. Nishimura, N. Kanzaki, M. Okamoto, H. Sawada, Y. Iiz-awa, M. Shiraishi, Y. Aramaki, K. Okonogi, Y. Ogawa, K. Meguro,M. Fujino, A small-molecule, nonpeptide CCR5 antagonist withhighly potent and selective anti-HIV-1 activity, Proc. Natl. Acad. Sci.USA 96 (1999) 5698–5703.

[20] K. Maeda, H. Nakata, Y. Koh, T. Miyakawa, H. Ogata, Y. Takaoka,S. Shibayama, K. Sagawa, D. Fukushima, J. Moravek, Y. Koyanagi,

H. Mitsuya, Spirodiketopiperazine-based CCR5 inhibitor which pre-serves CC–chemokine/CCR5 interactions and exerts potent activityagainst R5 human immunodeWciency virus type 1 in vitro, J. Virol. 78(2004) 8654–8662.

[21] I. Navratilova, M. Pancera, R.T. Wyatt, D.G. Myszka, A biosensor-based approach toward puriWcation and crystallization of G protein-coupled receptors, Anal. Biochem. In press.