Forskolin-free cAMP assay for G -coupled receptors Biochemical Pharmacology

Forskolin-free cAMP assay for G _i -coupled receptors

Julie Gilissen

^a,b

, Pierre Geubelle

^a

, Nadine Dupuis

^a

, Céline Laschet

^a

, Bernard Pirotte

^b

, Julien Hanson

^a,b,

*

aLaboratoryofMolecularPharmacology,GIGA-SignalTransductionUnit,UniversityofLiège,11,Avenuedel'hôpital,4000Liège,Belgium

bLaboratoryofMedicinalChemistry,CentreforInterdisciplinaryResearchonMedicines(CIRM),UniversityofLiège,15,AvenueHippocrate,4000Liège, Belgium

ARTICLE INFO

Articlehistory:

Received21July2015 Accepted11September2015 Availableonline16September2015

Keywords:

GPCR SUCNR1 cAMP Forskolin GPR91

ABSTRACT

Gprotein-coupledreceptors(GPCRs)representthemostsuccessfulreceptorfamilyfortreatinghuman diseases. Manyare poorlycharacterizedwithfew ligandsreportedor remaincompletelyorphans.

Therefore, thereis agrowingneedforscreening-compatibleand sensitiveassays.Measurementof intracellularcyclicAMP(cAMP)levelsisavalidatedstrategyformeasuringGPCRsactivation.However, agonistligandsforGi-coupledreceptorsaredifficulttotrackbecauseinducerssuchasforskolin(FSK) mustbeusedandaresourcesofvariationsanderrors.

WedevelopedamethodbasedontheGloSensorsystem,akineticassaythatconsistsinaluciferase fusedwithcAMPbindingdomain.Asaproofofconcept,weselectedthesuccinatereceptor1(SUCNR1or GPR91)whichcouldbeanattractivedrugtarget.Ithasneverbeenvalidatedassuchbecauseveryfew ligandshavebeendescribed.

FollowinganalysesofSUCNR1signalingpathways,weshowthattheGloSensorsystemallowsreal time, FSK-free detectionofanagonist effect. This FSK-freeagonistsignalwas confirmedonother Gi-coupledreceptorssuchasCXCR4.InatestscreeningonSUCNR1,wecomparedtheresultsobtained withaFSKvsFSK-freeprotocolandwereabletoidentifyagonistswithbothmethodsbutwithfewerfalse positiveswhenmeasuringthebasallevels.

Inthisreport,wevalidateacAMP-inducerfreemethodforthedetectionofGi-coupledreceptors agonistscompatiblewithhigh-throughputscreening.

ThismethodwillfacilitatethestudyandscreeningofGi-coupledreceptorsforactiveligands.

ã2015ElsevierInc.Allrightsreserved.

1.Introduction

G protein-coupled receptors (GPCRs) are characterized by seven transmembrane domains and represent the largest family of proteins in the human genome

[1].

They are currently the target for 30% of marketed drugs and thus the most successful receptor family for treating human diseases. However, among the 350 non- olfactory members, many are poorly characterized with few ligands reported or remain completely orphans (around 100 in the most recent IUPHAR list)

[2].

GPCRs signal through G proteins but

also many intracellular partners such as arrestins

[3].

There are four main families of G proteins: G

i/o

, G

s

, G

q/11

and G

12/13

, which differ in the signaling pathways they couple to

[4].

The ef

fi

cient identi

fi

cation of original ligands for unknown and poorly characterized receptors remains a major challenge. To reach this goal, a plethora of assays have been developed, in response to the high demand for ligands both for therapeutic and research perspective. More recently, it has been reported that some ligands could selectively activate discrete signaling pathways when binding to a receptor

[5].

This pharmacological property called functional selectivity is now being considered when selecting an assay for screening campaigns. Therefore, there is a renewed interest in screening-compatible and sensitive assays directed selectively toward a pathway of interest.

Measurement of intracellular cyclic adenosine monophosphate (cAMP) levels is a validated strategy for such pathway speci

fi

c approach

[6].

This prominent second messenger is the product of adenylate cyclase (AC) activity that is directly regulated by G

s

- and

Abbreviations:AC,adenylatecyclase;AUC,areaunderthecurve;cAMP,cyclic

adenosine monophosphate; ERK, extracellular signal regulated kinases; FSK, forskolin;GPCRs,G-proteincoupledreceptors;PTX,pertussistoxin;SA,succinic acid.

* Correspondingauthorat:LaboratoryofMolecularPharmacology,GIGA-Signal TransductionUnit,UniversityofLiège,CHU,B34,TourGIGA(+4),Avenuedel’hôpital, 11,4000Liège,Belgium.Tel.:+3243664748.

E-mailaddress:j.hanson@ulg.ac.be(J.Hanson).

http://dx.doi.org/10.1016/j.bcp.2015.09.010 0006-2952/ã2015ElsevierInc.Allrightsreserved.

ContentslistsavailableatScienceDirect

Biochemical Pharmacology

j o u r n a l h o m e p a g e : w w w . e l s e vi e r . c o m / l o c a t e/ b i o c h em p h a r m

G

i

-proteins. The vast majority of GPCR activation can be monitored with the changes in cAMP levels. The G

s

-coupled receptors activation is relatively easy and straightforward to detect because they activate AC and consequently increase cAMP levels. Accord- ingly, many examples of successful screening campaigns on G

s

-coupled receptors have been published

[7,8].

In contrast, agonist ligands for G

i

-coupled receptors are much more dif

fi

cult to track with cAMP measurement. This is due to the fact that basal AC activity and cAMP levels in the cell are relatively low

[9].

Inducers such as forskolin (FSK)

[10]

or similar stratagem

[11]

must be used when assessing a putative agonist. The arti

fi

cial manipulation of the signal complicates the assay by increasing the sources of variation and errors

[12].

In this report, we describe and validate a cAMP-inducer free method for the detection of G

i

-coupled receptors agonists compatible with high-throughput screening. The method is based on the GloSensor system, a live cell, homogenous and kinetic assay that consists in a luciferase fused with cAMP binding domain

[13].

As a proof of concept, we selected the succinate receptor 1 (SUCNR1, previously termed GPR91) that is coupled to G

i[14,15].

The receptor-ligand pair has been described as a metabolism sensor because succinic acid (SA) is a citric acid cycle intermediate that is released outside the cell in case of oxygen deprivation

[16].

A lot of studies have addressed the roles of SUCNR1 and demon- strated its implication in the enhancement of immunity

[17],

retinal angiogenesis

[18],

hypertension

[15,16,19],

liver damage

[20]

and platelet aggregation

[21,22].

Collectively, these data suggest that SUCNR1 could be an attractive drug target in several pathologies. However, no synthetic agonists and very few ligands have been described

[23].

Herein, we analyze SUCNR1 signaling pathways and show that SUCNR1 couples preferentially to G

i

. We further demonstrate that the GloSensor system allows FSK-free detection of an agonist effect. Moreover, we show that the performance of the assay was not modi

fi

ed by the addition of FSK. When we compared the results obtained in a test screening with a FSK vs. FSK-free protocol, we identi

fi

ed SUCNR1 agonists with both methods but with fewer false positive when measuring the basal levels.

2.Materialandmethods

2.1. Material

All chemicals used were from Sigma

–

Aldrich (St. Louis, Misouri, USA) unless otherwise stated. CXL12 (300-28A) was from PeproTech (Rocky Hill, New Jersey, USA). The following commer- cially available antibodies were used for several applications:

monoclonal anti-FLAG clone M2 (F3165) from Sigma

–

Aldrich (St.

Louis, Misouri, USA); anti-Mouse IgG (H + L), F(ab

⁰

)

2

Fragment (#4408, Alexa Fluor

¹

488 Conjugate) from Cell Signaling Technology (Danvers, Massachusetts, USA); rabbit monoclonal anti-phospho-p42/44 MAPK antibody (Thyr202/Thyr204, D13.14.4E) from Cell Signaling Technology (Danvers, Massachu- setts, USA); rabbit polyclonal IgG anti-Hsp90 a / b antibody (H-114) from Santa Cruz Biotechnology (Dallas, Texas, USA) and anti-rabbit IgG, HRP-linked antibody (#7074) from Cell Signaling Technology (Danvers, Massachusetts, USA).

2.2. Cell culture

Human embryonic kidney 293 (HEK293) cells were from American Type Culture Collection (ATCC, USA) and grown in DMEM adjusted to contain 10% fetal bovine serum (FBS, Biochrom AG, Berlin, Germany), 1% penicillin and streptomycin (Lonza, Verviers, Belgium), 1% L-glutamine (Lonza, Verviers, Belgium) at 5% CO

2

and 37

C.

SUCNR1 coding sequence was ampli

fi

ed from human genomic DNA and cloned into the pIRESpuro expression vector (Clontech Laboratories, Mountain View, California, USA).

The pGloSensor

^TM

-22F cAMP (cAMP GloSensor) plasmid was obtained from Promega Corporation (Madison, Wisconsin, USA).

Human arrestin 2 was ampli

fi

ed from cDNA prepared from HEK293 cells and cloned into the pIREShygro expression vector (Clontech Laboratories, Mountain View, California, USA). Arrestin 3 was ampli

fi

ed from b -arrestin 2 GFP WT (#35411, Addgene, Cambridge, Massachusetts, USA) and cloned into the pIREShygro expression vector. The pIRES.hygro.FnLARR2, pIRES.hygro.

FnLARR3, pIREShygroFnLARR2.pIRESpuroSUCNR1 and pIREShy- groFnLARR3.pIRESpuroSUCNR1 were developed based on Taka- kura et al.

[24].

Brie

fl

y, the 1-415

fi

rst amino acids of

fi

re

fl

y luciferase were fused with N-Arrestin2 or 3 (FN-Arr2 or 3) and the 413-549 amino acids were fused with C-SUCNR1 (SUCNR1-FC).

2.3. Flow cytometry analysis

Cells (2 10

⁵

cells per tube) were incubated with monoclonal ANTI-FLAG M2 (1:1000) for 45 min at 4

C. After wash, cells were incubated with anti-Mouse IgG (H + L), F(ab

⁰

)

2

Fragment (Alexa Fluor

¹

488 Conjugate; 1:1000) for 45 min at 4

C in the dark. Data were acquired on BD FACSCalibur 2 lasers (Becton Dickinson, New Jersey, USA) and analyzed with Cellquest pro. The gate on living cells was made using the SSC/FSC dot plot.

2.4. Immuno

fl

uorescence staining and confocal microscopy

HEK293 cells lines were grown on poly-(

D

-lysine)-treated glass coverslips (VWR, 20 20 mm) at 37

C in 5% CO

2

for 24 h. The cells were incubated on ice 1 h in HBSS (120 mM NaCl, 5.4 mM KCl, 0.8 mM MgSO

4

, 10 mM HEPES; pH 7.4; 10 mM glucose) containing ANTI-FLAG M2 (1:1000). After several washing steps, cells were incubated 10 min in HBSS at 37

C,

fi

xed for 5 min on ice and 15 min at room temperature (RT) in PBS containing 4% paraformaldehyde.

Cells were blocked and permeabilized at RT for 30 min with PBS containing 2% BSA and 0.12% Triton X-100. After wash, cells were incubated with PBS containing 2% BSA, 0.12% Triton X-100 and anti- Mouse IgG (H + L), F(ab

⁰

)

2

Fragment (Alexa Fluor

¹

488 Conjugate;

1:1000) for 1 h and 45 min at RT in the dark. Cells were washed and glass coverslips mounted on slides (Marien

fi

eld, Germany) with Prolong Gold Antifade reagent containing dapi (Thermo Fischer Scienti

fi

c/Life Technologies, Waltham, Massachusetts, USA).

Images were acquired using confocal microscope (Nikon A1R).

2.5. cAMP assay

HEK293 cells stably transfected with plasmid containing cAMP

GloSensor or pGlo and pIRESpuroSUCNR1 were selected with

hygromycin 200 m g mL

¹

(A.G. scienti

fi

c, San Diego, California,

USA) and puromycin 2 m g mL

¹

. Prior to the experiment, cells were

starved for 5 h with 1% FBS. Cells from a con

fl

uent T175

fl

ask were

detached and incubated 1 h in the dark at RT in assay buffer HBSS

(120 mM NaCl, 5.4 mM KCl, 0.8 mM MgSO

4

, 10 mM HEPES; pH 7.4,

10 mM glucose) containing IBMX (300 m M) and luciferin in HEPES

buffer (GloSensor reagent, Promega) according to manufacturer

instructions. Cells were distributed into 96-well plates

(150,000 cells per well or 37,000 cells per well in 384-well plates,

white Lumitrac

^TM

, Greiner) containing the ligands at different

concentrations. After 1 min agitation at 1200 rpm and 9 min

incubation with compounds, basal luminescence level was

recorded. Similarly, luminescence was recorded following injec-

tion of FSK (40 measures; 500 ms integration time). The

luminometer was a Fluoroskan Ascent FL plate reader (Thermo

Electron Corp., ascent software version 2.6) equipped with

2 dispensers. In the experiment with PTX, cells were incubated overnight with 100 ng mL

¹

of PTX (Calbiochem/Merck Millipore, USA) prior to assay.

2.6. Intracellular calcium mobilization assay

The assay has been conducted according to previous description

[25].

Brie

fl

y, cells from a con

fl

uent T175

fl

ask were detached and

incubated in assay buffer (HBSS: 120 mM NaCl, 5.4 mM KCl, 0.8 mM MgSO

4

, 10 mM HEPES; pH 7.4, 10 mM glucose) containing 5 m M coelenterazine h (regis technologies, USA) for 1 h in the dark at 37

C. Before stimulation with ligands, the coelenterazine-con- taining buffer was replaced by assay buffer supplemented with 1.8 mM CaCl

2

. Luminescence was followed for 8 s (40 measures;

200 ms integration) immediately upon ligand addition. Measure- ments were acquired with a Fluoroskan Ascent FL (Thermo

Fig.1.Celllinecharacterization,arrestinbindingand[Ca²⁺]imobilization.

(A)Cell-surfacereceptorexpressionanalyzedbyflowcytometryonHEK293.pGlo.SUCNR1cells(notlabeledingreycomparedtoFlaglabeledinwhite).(B)SUCNR1is internalizedinaconstitutivemannerinHEK293cellsexpressingFlag-taggedSUCNR1at37C(right)compareto0C(left).(C)SUCNR1isabletorecruitarrestin3 (EC50>1mM)andarrestin2(EC50>2mM)whenstimulatedwithSAinARR3.SUCNR1andARR2.SUCNR1cells,respectively;SAhasnoeffectonMocktransfectedcells.(D) CalciummobilizationinducedbySA(EC50=292.90.9mM)inHEK293.G5A.SUCNR1cellsusinganaequorinassay.PretreatmentwithPTXcompletelyabolishedthecalcium mobilizationinducedbySA.DataareexpressedasmeanSEMofatleast3independentexperiments.(E)ActivationofERK1/2bystimulationofuntreatedandPTX-treated HEK293.pGlo.SUCNR1cellswithSAat500mMduring3min.Shownisrepresentativeofatleast4experiments.

Electron Corp., ascent software version 2.6, equipped with 2 dispensers).

2.7. Arrestin complementation assay

HEK293 cells stably transfected with pIREShygroFnLARR2 or pIREShygroFnLARR3 and pIRESpuroSUCNR1 were selected with hygromycin 200 m g mL

¹

(or 400 m g mL

¹

) and puromycin 1 m g mL

¹

. Cells suspension (HBSS with 20 mM HEPES, pH 7.4, 10 mM glucose) were incubated into 96-well plates (100,000 cells per well) containing the ligands at different concentrations for 10 min at RT. Following injection of 50 m M luciferin (Synchem, Germany), luminescence was recorded for 30 min using a high sensitivity luminometer (Berthold technologies, Centro XS

³

LB 960, MicroWin 2000 software, equipped with 2 dispensers).

2.8. Determination of ERK phosphorylation

HEK293 cells stably transfected with pIRESpuroSUCNR1 (selected with puromycin 1 m g mL

¹

) were plated in 6-well plates, starved with 1% FBS and pretreated with 100 ng mL

¹

PTX or vehicle overnight. Cells were incubated with succinic acid for 3 min

at 37

C. Cells were immediately put on ice, lysed with ice-cold RIPA Buffer (25 mM Tris

–

HCl, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS; pH 7.6) supplemented with protease inhibitors and phosphatase inhibitors (Roche, Basel, Switzerland).

Cell lysate were analyzed by SDS page electrophoresis followed by immunoblotting. ERK1/2 phosphorylation was detected with a rabbit monoclonal anti-phospho-p42/44 MAPK antibody (Thyr202/Thyr204, D13.14.4E, Cell Signaling, 1:2000), and Hsp90 was detected with a rabbit polyclonal IgG anti-Hsp90 a / b antibody (Santa Cruz Biotechnology, 1:5000). The membranes were then probed with corresponding HRP-conjugated secondary antibody (Cell Signaling Technologies, 1:2000).

2.9. Data analysis and statistical procedure

All data analyses were performed using computer software (GraphPad Prism version 5.0 for Windows).

Statistical analyses of differences between 2 groups were performed by non-parametric, unpaired, 2-tailed Mann

–

Whitney test. P values less than 0.05 were considered as statistically signi

fi

cant.

B

C D

A

0 20 40 60 80 100

+ - + - + -

- + - + - +

- - - - + +

MOCK SUCNR1

*

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

0 50 100 150 200 250 pGlo

MOCK

Log[Isoproterenol]

-8 -7 -6 -5 -4

0 50 100 150 200 250 pGlo

MOCK

Log[FSK]

-7 -6 -5 -4 -3

0 10 20 30

SUCNR1-PTX SUCNR1+PTX MOCK

Vehicle SA500µM

PTX

Log[SuccinicAcid]

R.L.U. (% of ctrl) R.L.U. (% of ctrl)

R.L.U. (% of ctrl)

300 300

cAMP inhibition (% of ctrl)

Fig.2.cAMPinhibitionmediatedbySUCNR1activation.

(A)HEK293cellsstablytransfectedwiththeGloSensorcAMPbiosensor(HEK293.pGlo)showconcentration-dependentincreaseinRelativeLuminescentUnits(R.L.U.)when treatedwithincreasingconcentrationsoftheadenylatecyclaseactivatorforskolin(EC50=870.589.0nM).(B)Effectofisoproterenol,adrenoceptorsagonist,oncAMP productionusingHEK293.pGlocells(EC50=153.64.1nM).(C)EndpointmeasureoftheeffectofSAat500mMonintracellularcAMPstimulatedwith1mMforskolinusing HEK293.pGlo.SUCNR1cells(n=3;p<0.05).(D)SAdecreasescAMPlevelsstimulatedwithforskolin1mMinHEK293.pGlo.SUCNR1cellsinaPTX-sensitiveandconcentration dependentmanner(EC50=790.1mM).DataareexpressedasmeanSEMofatleast3experiments.

2.10. Calculation of Z

⁰

factor

Z

⁰

values were determined to monitor assay quality and were calculated according to the formula:

Z

⁰

= 1 ((3 s

c+

+ 3 s

c

)/| m

c+

m

c

|)

[26].

2.11. Hit selection and activity cut-off criteria

We set up two criteria for hit selection: (1) a positive activity on SUCNR1 expressing cells > negative control (vehicle) mean + 6 s

and (2) activity on Mock cells (HEK293.pGlo) comprised within negative control (vehicle) mean 3 s ^{. Compounds ful}

fi

lling these 2 criteria were selected for secondary screening. The number of compounds selected represented approximately 0.15% of the collection (Hit rate). Following cherry picking, compounds were assayed in triplicate at one concentration on SUCNR1 and mock cells. Compounds showing statistically signi

fi

cant activity on SUCNR1 were selected for complete concentration

–

response curves.

3.Results

3.1. Cell line characterization, arrestin binding and [Ca

²⁺

]

i

mobilization We generated HEK293 cell line stably expressing N-terminus

fl

ag tagged SUCNR1 and veri

fi

ed its expression at the cell membrane by FACS analysis (Fig.

1A)

and its ability to internalize in a constitutive manner at 37

C (Fig.

1B).

Since arrestins are reported as being responsible for GPCRs internalization

[27],

we analyzed the capacity of activated SUCNR1 to recruit arrestin 2 and 3. Using a protein complementation strategy validated with the

b

2

-adrenoceptor (data not shown), we showed that arrestin 3 (EC

₅₀

> 1 mM) and arrestin 2 (EC

₅₀

> 2 mM) could be recruited upon SA binding (Fig.

1C).

These results were consistent with the literature

[15,28].

We further con

fi

rmed the ability of SA to induce [Ca

²⁺

]

i

mobilization (Fig. 1D, EC

50

= 292.9 0.9 m M) in an aequorin- based assay. The signal was abolished when the cells were preincubated overnight with pertussis toxin (PTX, 100 ng ml

¹

). In addition, we detected an increase in phosphorylated extracellular signal-regulated kinases (ERK) in HEK293.SUCNR1 cells upon SA addition (Fig.

1E).

This response was not detectable following PTX overnight incubation.

3.2. cAMP inhibition mediated by SUCNR1 activation

We set up a GloSensor cAMP bioassay by stably transfecting the plasmid pGloSensor-22F

^TM

into HEK293 cells. We determined the EC

50

of different cAMP inducers in order to determine the concentrations that should be used in our system. We could detect a robust signal with increasing levels of FSK (EC

50

= 870.5 89.0 nM) and isoproterenol (EC

50

= 153.6 4.1 nM), a potent agonist for endogenous adrenoceptors coupled to G

s

in HEK293 cells (Fig.

2A

and B). We used this HEK293.pGlo cell line to stably transfect SUCNR1 subcloned in a bicistronic IRES vector allowing the simultaneous expression of two proteins from the same transcript. The generated cell line HEK293.pGlo.SUCNR1 displayed a stable expression of the receptor over time, when the cells were grown in a selection medium. We validated the cell line by performing end-point assays in the presence of FSK used at the concentration of 1 m M, the approximate EC

50

in our system as determined by titration experiment (Fig.

2A).

Incubation of SA (500 m M) and FSK (1 m M) for 40 min resulted in a 28.3 1.4%

decrease compared to control upon activation of SUCNR1 (Fig.

2C).

Complete concentration

–

response curves permitted the calcula- tion of an EC

50

of 79 0.1 m M, consistent with published literature

(Fig.

2D)[15].

Cells preincubated with PTX or devoid of receptor did not respond to SA (Fig.

2C

and D).

3.3. Real time analysis of cAMP levels modulation mediated by SUCNR1 activation

The GloSensor system is compatible with kinetic measurement

[13]

and we were able to follow the evolution of signal upon addition of FSK (Fig.

3A).

The effect of SA was stable over time during the experiment (40 min) and already visible at 10 m M (Fig.

3A),

as expected from the concentration-response curve (Fig.

2D).

We observed that the levels of cAMP were already decreased in the presence of SA at the

fi

rst measure. This effect was concentration-dependent and we reasoned that it could be the direct effect of SA on the system. In order to validate this hypothesis, we followed the basal levels of cAMP for 30 min and injected SA at the concentration of 500 m M (Fig.

3B).

Although the signal was stable before the addition of SA, it immediately dropped further below the baseline (Fig.

3B).

We reversed the experiment and analyzed the effect of the addition of the agonist after the injection of FSK (Fig.

3C).

The signal induced by FSK was inhibited by SA although the level did not go back to basal but reached a plateau (Fig.

3C).

When we measured the integration over time (Area under the curve or AUC) for 5 min on basal levels (Fig.

3D,

grey bars) or for 40 min post-addition of FSK (Fig.

3D,

black bars), the effect of SA reached signi

fi

cance in both conditions (p

<

0.01).

We further con

fi

rmed the activity of SA on basal cAMP levels through SUCNR1 activation with the determination of a complete concentration

–

response curve (Fig.

3E).

We calculated an EC

50

= 22.83 0.03 m M for SA decrease of basal cAMP levels and an EC

50

= 45.79 0.08 m M for the inhibitory effect of SA on FSK induced cAMP (Fig.

3E)

that were signi

fi

cantly different (p

<

0.05).

Interestingly, the E

max

(E

max

= 52.3 2.7% of control) obtained on cAMP basal level was signi

fi

cantly (p

<

0.05) greater than maximal inhibition in the presence of FSK (E

max

= 38.0 1.5% of control). We investigated a range of FSK concentrations (0.1

–

1 m M) and observed that SA EC

50

and E

max

were dependent on FSK concentration (Fig.

3F).

In order to exclude that the effect on cAMP basal levels was limited to SUCNR1 or heterologously expressed receptors, we determined agonist potency of CXCL12 on endogenous CXCR4

[29,30]

with the same methodology (Fig.

3G).

We calculated an EC

50

= 16.01 1.07 nM and E

max

= 62.9 0.2 (% of control) for CXCL12-induced decrease of basal cAMP levels and an EC

50

= 13.93 1.12 nM and E

max

= 16.9 2.4 (% of control) for the inhibitory effect of CXCL12 on FSK-induced cAMP production. The differences between E

max

were statistically signi

fi

cant (p

<

0.0007).

The observed differences in EC

₅₀

and E

_max

may be partially explained by variations in luciferin concentration in assay buffer, before and after FSK injection. We tested several assay buffers with different luciferin content but did not see any effect on SA response (data not shown).

3.4. Optimization of a screening protocol and assay performance

We designed a protocol compatible with the screening of

chemical libraries (Fig.

4A).

In addition, we wanted to compare the

effects of compounds on basal and FSK-induced cAMP levels. 1 m l

of drug solutions from the library were distributed in 96-well

plates. 100 m l of a cell suspension was added and mixed

thoroughly. The mixture between drugs and cells was incubated

at RT for 10 min (Fig.

4A).

The basal level of each well was measured

for 5 min and the AUC was determined. These results constituted

the

“

basal level screen

”

, pictured in gray in

Fig.4A.

Forskolin at

1 m M

fi

nal concentration was added to each well and a kinetic

measurement was taken for 40 min. The signal was integrated and

B

C D

A

Drug Addition

Drug Addition FSK

Addition

E F

****

**

G

cAMP levels (R.L.U.)

3.5

2.0 2.5 3.0

1.5

0.0 0.5 1.0

0 400 800 1200 1600

Time (s)

0.4 0.5

0.3

0.0 0.1 0.2

0 1000 2000 3000 3850

Time (s)

Vehicle SA 10µM SA 500µM

SA 500µM Vehicle

Vehicle SA 500µM

FSK 1µM

cAMP levels (AUC, R.L.U.)

2.0 2.5

1.5

0.0 0.5 1.0

2000 2500

1500

0 500 1000

0 1000 2000 3000 3850

Time (s)

Log [Succinic Acid] Log [Succinic Acid]

Log [CXCL12]

- FSK + FSK

cAMP inhibition (% of ctrl) cAMP inhibition (% of ctrl)

40 50

30

0 10 20

80

60

0 20 40

FSK 0.1µM FSK 0.5µM FSK 1µM Vehicle

+ - + -

- + - +

- - + +

Vehicle

SA 500µM

- FSK + FSK + PTX 60

0 20 40

cAMP inhibition (% of ctrl)

-8 -7 -6 -5 -4 -3 -7 -6 -5 -4 -3

-8 -7

-9

cAMP levels (R.L.U.)

cAMP levels (R.L.U.)

Fig.3.RealtimeanalysisofcAMPlevelsmodulationmediatedbySUCNR1activation.

(A)SAeffectisstableoverthetimeofexperiment(40min)andalreadyvisibleat10mM.Atthefirstmeasure,cAMPlevelswerealreadybelowcontrolbaseline.(B)BasalcAMP levelsdroppedbelowthebaselineafterinjectionofSAat500mM.(C)cAMPlevelsinducedby1mMforskolinareinhibitedbySAat500mMalthoughtheleveldidnotgoback tobasalbutreachedaplateau.(D)Comparisonoftheintegrationovertime(expressedasareaunderthecurveorAUC)onbasallevels(whitebars)orfor40minpost-addition ofFSK(blackbars)inpresenceofSAat500mM(p<0.01).(E)Concentration–responsecurveforSAonHEK293.pGlo.SUCNR1cellsstimulatedwithforskolin1mM (EC50=45.790.08mM;Emax=38.01.5%comparedtocontrol)ornotstimulated(EC50=22.830.03mM;Emax=52.32.7%ofcontrol).(F)Concentration–responsecurve

expressed as AUC for each well. These results were called the

“

FSK- induced screen

”

, pictured in black, in

Fig.4A.

First, we determined the assay performance for the two different measurements in 96- wells plates. Basal cAMP level values distribution is given in

Fig.4B.

We determined the Z

⁰

factor to be 0.81. The values obtained following FSK stimulation are shown in

Fig.4C

and the calculated Z

⁰

factor was 0.74. We determined the assay performance with the same procedure on 384-wells plates and found a Z

⁰

factor = 0.75 for

B

C D

A

Drug Addition

FSK Addition

0 6 12 18 24 30 36 42 48 0

2000 4000 6000 8000 10000 12000

0 6 12 18 24 30 36 42 48

0 40 80 120 160 200 240

0 24 48 72 96 120 144 168 192 0

25 50 75 100 125

E

0 24 48 72 96 120 144 168 192 0

500 1000 1500 2000 2500 3000 3500

cAMP level (AUC, R.L.U.) cAMP level (R.L.U.) cAMP level (AUC, R.L.U.) cAMP level (AUC, R.L.U.)

cAMP level (AUC, R.L.U.)

Well # Well #

Well # Time (min)

Vehicle

SA 500µM

Vehicle

SA 500µM

Vehicle

SA 500µM

Well #

Vehicle

SA 500µM

Vehicle

SA 500µM Z’=0.61

Z’=0.75 Z’=0.81

Z’=0.74 0 5 1015 2025303540 4550 0

2 4 6 8

Fig.4.Optimizationofascreeningprotocolandassayperformance.

DesignofascreeningprotocolfordirectcomparisonbetweentheeffectofcompoundsonbasalandFSK-inducedcAMPlevels.(A)Themixturebetweendrugsandcellsis incubatedatRTfor10min.First,the“basallevelscreen”(inlightgray)isperformedandexpressedasAUCofbasallevelofeachwellmeasuredfor5min.Next,theforskolinat 1mMfinalconcentrationisaddedtoeachwellandakineticmeasurementistakenfor40min.ThesignalisintegratedandexpressedasAUCforeachwell.Theseresultsare calledthe“FSK-inducedscreen”(inblack).(B)Assayperformanceforthetwodifferentmeasurementswith96-wellsplates:Z⁰factorcalculatedforbasallevelmeasurementis Z⁰=0.81(C)andafterFSKstimulationisZ⁰=0.74.(D)Z⁰factorforbasallevelmeasurementhasbeencalculatedfor384-wellsplatestobeZ⁰=0.75and(E)Z⁰=0.61afterFSK stimulation.

forSAonHEK293.pGlo.SUCNR1cellsstimulatedwithdifferentconcentrationsofFSK.IncreasedconcentrationsofFSKinducedincreasedEC50anddecreasedEmaxforSA (EC50=18.864.33mM; Emax=74.11.7% of control without FSK; EC50=26.965.31mM; Emax=72.32.3% of control for 0.1mM FSK; EC50=41.7310.5mM;

Emax=58.285.9%ofcontrolfor0.5mMFSK;EC50=26.965.31mM;Emax=54.879.1%ofcontrolfor1mMFSK).(G)Concentration–responsecurveforCXCL12on HEK293.pGlocellsstimulatedwithforskolin1mM(EC50=13.931.12nM;Emax=16.92.4%comparedtocontrol)ornotstimulated(EC50=16.011.07nM;Emax=62.9 0.2%ofcontrol).PretreatmentwithPTXcompletelyabolishedtheeffectofCXCL12onHEK293.pGlocells.DataareexpressedasmeanSEMofatleast3experiments.

SA inhibition of basal cAMP levels (Fig.

4D)

and Z

⁰

factor = 0.61 (Fig.

4E)

for SA inhibition of FSK-induced cAMP production.

3.5. Screening of the Sigma LOPAC

^1280TM

library

We applied our protocol on a test screening of the Sigma LOPAC

^1280TM

library, constituted by 1280 compounds of known activity

[31]

distributed in sixteen 96-well plates. The compounds were diluted to give a

fi

nal concentration of 100 m M. We performed the screening on the HEK293.pGlo.SUCNR1 and a counter-screening on the HEK293.pGlo cell line. As expected, many compounds had an in

fl

uence on both the basal and FSK-induced cAMP levels. Therefore, we distributed the results on 2 axis: y axis representing the effect (% of inhibition) of compounds on SUCNR1 and x axis representing the effect (% of inhibition) on HEK293.pGlo cells (Fig.

5A).Fig. 5A

presents the plotted values of the basal measurements. We set two thresholds for the selection of hits: an inhibition above mean + 6 s intra-plate on SUCNR1 cell line and an activity comprised within mean 3 s intra-plates on HEK293.pGlo cells. For the

“

basal level screen

”

, 30 compounds met the criteria whereas 48 compounds were selected in the

“

FSK-induced screen

”

. 11 compounds were common to the two sets (Fig.

5B).

3.6. Secondary screening of the selected hits

We performed a cherry pick of the compounds that met our criteria in both conditions (67 compounds) and tested them according to the protocol depicted in

Fig.4A,

in triplicate.

Fig.6A

shows the results for the compounds that displayed a statistically signi

fi

cant difference between activity on HEK293.pGlo.SUCNR1 and HEK293.pGlo cells. All the 13 con

fi

rmed active compounds had succinic acid or maleic acid (a weaker SUCNR1 agonist

[15])

as a

counter ion (Table

1).

We reasoned that we detected an agonist activity because of the presence of the counter ion. We bought some of the compounds in another chemical form to con

fi

rm this hypothesis. For instance, compound 3E8 in the library, BRL 54443, a 5-hydroxytryptamine receptor 5-HT

1e

& 5-HT

1F

agonist

[32],

showed no activity alone in our assay (Fig.

6B)

whereas maleic acid con

fi

rmed its activity and showed an EC

50

= 93.8 1.3 m M and E

max

= 32.6 3% when assayed in the presence of FSK. Its EC

50

on basal cAMP levels was 79.4 1.1 m M and E

_max

= 49.4 3.9%

compared to control (Fig.

6C).

7 compounds in the set of 13 con

fi

rmed hits (3E2, 5D3, 11A4, 11C4, 11H3, 11H6 and 15G10) were identi

fi

ed in the

“

basal level screen

”

but remained unnoticed in the

“

FSK-induced screen

”

.

4.Discussionandconclusions

In the present study, we demonstrated the feasibility of assaying and detecting agonist ligands of G

i

-coupled receptors directly from the inhibition of the basal cellular levels of cAMP.

Bioassays for GPCRs cover a wide range of strategies from speci

fi

c biosensor for second messenger to binding assays.

Recently, increased attention has been given to pharmacological assays because of the novel paradigm that some ligand-receptor pairs may display functional selectivity that translate in com- pounds activating only discrete signaling pathways

[5].

Therefore, unbiased approaches such as label-free or arresting binding assays have been widely publicized as being a superior set up for the identi

fi

cation of GPCR ligands

[33,34].

In theory, they can detect the binding of any activating ligand with a universal read-out. The advantages of the approach being that the wider the net, the more substances might be identi

fi

ed

[34].

While there is an evident interest for truly general assays, some receptors have been

A B

Basal level (-FSK) FSK-Induced Common

Inhibition of cAMP levels HEK293.pGlo.SUCNR1 (% of ctrl) 100

-100 -80 -60 -40 -20 0 20 40 60 80

100

-5 0

-10 -15 -20 0 20 40 60 80

-25 5 10 15 20 25

-20 0 -40

-60 -80

-100 20 40 60 80 100

Inhibition of cAMP levels HEK293.pGlo.SUCNR1 (% of ctrl)

Inhibition of cAMP levels HEK293.pGlo (% of ctrl)

Inhibition of cAMP levels HEK293.pGlo (% of ctrl)

Fig.5.ScreeningoftheSigmaLOPAC^1280TMlibrary.

(A)Theligandswerescreenedat100mMoncellsexpressingSUCNR1andcounter-screenedonmockcelllinetoestimatetheireffectoncAMPlevelsintheabsenceofSUCNR1.

Compoundswithanactivity(%ofinhibition)superiortomean+6s^{comparedtovehicleon}HEK293.pGlo.SUCNR1cellsandanactivitycomprisedbetweenmean3s^on HEK293.pGlocellswereselectedtobeevaluatedinasecondaryscreening.(B)Wecomparedtheresultsobtainedforthe“basallevelscreen”(30compoundsidentifiedas invertedtriangle)andthe“FSK-inducedscreen”(48compoundsidentifiedbysquares),11compoundscommontothetwosetsarerepresentedasfilledcircles.

described as being not, or not well, coupled to arrestins or other canonical pathways. For instance, prominent examples of GPCRs lacking the ability to bind arrestins include the b

3

-adrenoceptor, the relaxin family peptide receptor 1 (RXFP1) and 2 (RXFP2), or the glucose-dependent insulinotropic polypeptide (GIP) receptor

[35–38].

More surprisingly, some GPCRs such as the atypical chemokine receptor ACKR3 or receptors involved in stem cells biology (LGR5) seem to lack G-protein-coupling

[39,40].

Therefore, a general paradigm in terms of signaling is dif

fi

cult to apply to the entire family or even sub groups of GPCRs. A

fi

t for purpose approach could increase the odds of identifying good ligands for poorly characterized receptors. In fact, it could be speculated that the growing observations that all GPCRs do not

fi

t into general signaling paradigm might explain why some receptors are still orphan, even of surrogate ligands.

We selected SUCNR1 to develop a screening assay focused on cAMP. The signaling pathways for this receptor remained somehow elusive with some discrepancies in the literature.

SUCNR1 was originally described as coupled to both G

i

and G

q

and to be internalized upon SA exposure in HEK293 cells

[15].

The view of SUCNR1 being coupled to G

q

has been later challenged.

Sundstrom et al. proposed that the [Ca

²⁺

]

_i

mobilization was a consequence of PLC- b activation by the dimer G bg

[41].

Our observation that SA elicits a concentration-dependent PTX- sensitive [Ca

²⁺

]

i

mobilization is consistent with SUCNR1 being not coupled to G

q

, at least when heterologously expressed in HEK293 cells. Using an in-house luciferase complementation assay 0

10 20 30 40 50

+ FSK - FSK

Log [BRL 54443]

0 10 20 30 40 50

+ FSK - FSK MOCK

Log [Maleic Acid]

SA 3E 2 3E 8

3H 3 5C 9

5D 3 6A 9

10 H5 11 A4

11 B 4 11 C 4

11 H3 11 H6

15G1 0 0

10 20 30 40 50

- FSK + FSK

A

B C

cAMP inhibition (% of ctrl) cAMP inhibition (% of ctrl)

cAMP inhibition (% of ctrl)

-8 -7 -6 -5 -4 -3

-3 -4 -5 -6 -7

Fig.6.Secondaryscreeningoncompoundsmeetingthecriteria.

67compoundsselectedwiththeprimaryscreeningwereevaluatedat100mMintriplicates.CompoundswithnosignificantactivityonHEK293.pGlo.SUCNR1cellscompareto HEK293.pGlocellsweredesignatedasfalsepositive.(A)Onlycompoundswithsuccinicacidormaleicacidascounterionsshowedsignificantactivityeitherwithadditionof forskolinornotonSUCNR1.(B)BRL54443withoutmaleateasacounterionisinactiveonSUCNR1.(C)Concentration–responsecurveofmaleicacidonSUCNR1cells stimulatedwithforskolin1mM(93.81.3mM;Emax=32.63%comparedtocontrol)ornotstimulated(EC50=79.41.1mM;Emax=49.43.9%comparedtocontrol).Data areexpressedasmeanSEMofatleast3independentexperiments.

Table1 Confirmedhits.

3E2 ()-Brompheniraminemaleate

3E8 BRL54443maleate

3H3 (+)-Brompheniraminemaleate

5C9 Doxylaminesuccinate

5D3 5-Carboxamidotryptaminemaleate

6A9 N,N-Dipropyl-5-carboxamidotryptaminemaleate

10H5 Methylergonovinemaleate

11A4 ()-MK-801hydrogenmaleate

11B4 2-Methyl-5-hydroxytryptaminemaleate

11C4 Alpha-methylserotoninmaleate

11H3 Dizocilipinemaleate

11H6 Nomifensinemaleate

15G10 S()-Timololmaleate

for arrestin recruitment detection, we show that activated SUCNR1 is able to couple to arrestin 3 and arrestin 2. However the receptor seems to be dramatically less ef

fi

ciently coupled to arrestin 2 (EC

50

> 2 mM) and 3 (EC

50

> 1 mM) compared to G

i

(EC

50

= 22.83 0.03 m M), although the level of receptor membrane expression was similar (data not shown). Indeed, we were not able to calculate the EC

50

for arrestin recruitment with this assay because higher concentrations induced non-speci

fi

c effects due to acidic nature of SA. This weak coupling is consistent with other reports for arrestin 3

[28]

and has at least two consequences. First, it could be postulated that arrestin doesn't play a signi

fi

cant physiological role in this receptor-ligand system since SA would have to reach important concentration (above 1 mM) to weakly activate this pathway. Secondly, for this kind of receptor, a screening based on arrestin assay would not be a good option. Ligands slightly less effective than SA would probably never be detected. Therefore, SUCNR1 can be considered as a good candidate for a G

i

-based screening assay.

G

i

-coupled receptor activity can be assessed by many techni- ques. Historically, GTP- g -S assays were commonly used but dif

fi

cultly amenable to large screenings due to the use of radioactive reagents

[42].

More recently, promiscuous G proteins have been utilized to couple G

i

-coupled receptors to phospholipase C- b (PLC- b ) and detect activation with [Ca

²⁺

]

i

transient mobiliza- tion. Actually, SA has been paired with SUCNR1 using this kind of approach

[15].

Although this method has proven its effectiveness in some deorphanization campaigns

[1],

it suffers from important drawbacks. Firstly, some special equipment such as FLIPR is required for signal acquisition and puts the technique (in a screening setting) out of reach for most academic labs or small sized companies. Secondly, the promiscuous G protein is another surrogate that does not effectively couple to all G

i

receptors or may modify the pharmacology of ligands

[43].

Another approach is the direct measurement of cAMP that has many advantages. The most obvious one being that it is the endogenous second messenger for G

i

and G

s

receptors, leaving the pharmacology and coupling system of the receptor unaffected. In addition, the signal is stable over time, compared to transient [Ca

²⁺

]

_i

mobilization. Several end point assays have been adapted for the direct or indirect detection of cAMP in cell lysate, with some variations in protocol and sensitivity

[6].

These assays all rely on competition between endogenous cAMP and some added labeled cAMP, which is a major disadvan- tage. In addition, they are not compatible with real time kinetic measure

[12].

Reporter gene assays based on cAMP sensitive transcription factor (CRE) activation are also available and frequently used

[12].

Recently, biosensors have been developed, opening the possibility of analyzing cAMP

fl

uctuations in living cells in a real time fashion

[12,13].

Biosensors can be considered superior compared to previous techniques on several aspects: they are more sensitive, no lysis is required and kinetic measurements on living cells are feasible.

The GloSensor system is such a biosensor that has been developed and marketed in 2008 by Promega Corporation

[13].

It has been extensively reported in the literature for various uses, from ligand identi

fi

cations for G

s

-coupled receptors

[44]

to the dissection of subtle pharmacological aspects of cAMP regulation such as biased signaling or endosomal cAMP generation

[45].

An improved version with increased dynamic range called

“

22F

”

(used in this study) was released a couple of years after the initial report

[46].

It was suggested by the authors that the sensitivity of the enhanced construct could be suf

fi

cient for recording cAMP inhibition of basal levels without FSK

[46].

However, to the best of our knowledge, it is the

fi

rst time that the ability of this assay to measure G

_i

mediated decrease of basal levels is comprehensively investigated. Without the use of FSK, we were able to monitor agonist activity on heterologously expressed SUCNR1 but also

endogenous CXCR4, a G

i

-coupled chemokine receptor expressed in HEK293 cells

[29].

It can be hypothesized that technologies previously lacked the required sensitivity and/or dynamic re- sponse to robustly detect variations in G

i

induced decrease of cAMP levels. During the preparation of this manuscript, we could

fi

nd only few reports of ligands identi

fi

ed with cAMP-based assays on G

i

receptors

[47],

and never for direct agonists. This is consistent with technical limitations that preclude the choice of a cAMP- based assay for G

_i

receptors, despite its advantages.

Although no original agonists were identi

fi

ed during the test screening, this FSK-free method was able to robustly identify true agonists of the receptor disseminated in the library as counter ions, despite an important proportion of compounds interfering with the system. It was rather unexpected to observe that FSK did not bring any improvement in our assays, in terms of EC

50

, E

max

or number of con

fi

rmed hits. However, the important artifacts that FSK can bring are well described, such as its impact on the activation of AC via G

s

-coupled receptors

[48,49].

In addition, it has poor aqueous solubility and the concentration to use must be determined since its EC

50

depends on the cell type and assay employed

[11].

Therefore, the choice of the working concentration will affect the potency of active compounds and may hide weakly active compounds or partial agonists resulting in false negatives

[11].

A con

fi

rmation of this assumption is the detection of 7 hits in our

“

basal level screen

”

that were not identi

fi

ed when using FSK.

These hits can be considered as false negative generated by FSK.

In summary, we established a screening-compatible FSK-free cAMP assay that allows the identi

fi

cation of agonist ligands for G

i

-coupled receptors. We selected SUCNR1 as a proof of concept given its elusive characterization. For the

fi

rst time, we report a cAMP level determination method, compatible with high-through- put screening that does not require the use of a cAMP inducer with G

i

-coupled receptor. Our protocol is readily available, easy to set up, fast and relatively cheap. Therefore, it should facilitate screening campaigns for G

i

-coupled receptors, especially for academic labs and small sized biotech companies that study G

i

receptors and do not have access to the [Ca

²⁺

]

i

-FLIPR assay.

Facilitating screening of G

_i

pathway brings also renewed oppor- tunities to screen G

i

-exclusive receptors that are unable to ef

fi

ciently couple to promiscuous G proteins and arrestins.

Conflictofinterest

The authors declare no con

fl

ict of interest.

Authorcontributions

JH designed and supervised the study. JG, PG, ND, CL and JH performed the experiments and acquired the data. JG, JH and BP analyzed the data, interpreted the results and wrote the paper.

Acknowledgments

This work was supported by the Fonds pour la Recherche Scienti

fi

que (F.R.S.-FNRS) Incentive Grant for Scienti

fi

c Research (F.4510.14), University of Liège (Crédit de Démarrage-Fonds Spéciaux) and Léon Fredericq Foundation. JH and CL are F.R.S.- FNRS Research Associate and Ph.D. fellow, respectively. ND is a FRIA PhD fellow. JG received grant from Léon Fredericq Foundation.

We thank the GIGA Imaging Platform for technical support in confocal image acquisition and FACS analysis. The luminometer (Berthold technologies, Centro XS

³

LB 960, MicroWin 2000 software, equipped with 2 dispensers) was provided by JC Twizere.

The authors gratefully acknowledge the technical assistance of C.

Piron.

References

[1]O.Civelli,R.K.Reinscheid,Y.Zhang,Z.Wang,R.Fredriksson,H.B.Schioth,G protein-coupledreceptordeorphanizations,Annu.Rev.Pharmacol.Toxicol.53 (2013)127–146.

[2]A.P.Davenport,S.P.Alexander,J.L.Sharman,A.J.Pawson,H.E.Benson,A.E.

Monaghan,etal.,Internationalunionofbasicandclinicalpharmacology.

LXXXVIII.Gprotein-coupledreceptorlist:recommendationsfornewpairings withcognateligands,Pharmacol.Rev.65(2013)967–986.

[3]A.C.Magalhaes,H.Dunn,S.S.Ferguson,RegulationofGPCRactivity,trafficking andlocalizationbyGPCR-interactingproteins,Br.J.Pharmacol.165(2012) 1717–1736.

[4]N.Wettschureck,S.Offermanns,MammalianGproteinsandtheircelltype specificfunctions,Physiol.Rev.85(2005)1159–1204.

[5]T.Kenakin,Functionalselectivityandbiasedreceptorsignaling,J.Pharmacol.

Exp.Ther.336(2011)296–302.

[6]C.Williams,cAMPdetectionmethodsinHTS:selectingthebestfromtherest, Nat.Rev.DrugDiscov.3(2004)125–135.

[7]C.Z.Chen,N.Southall,J.Xiao,J.J.Marugan,M.Ferrer,X.Hu,etal.,Identification ofsmall-moleculeagonistsofhumanrelaxinfamilyreceptor1(RXFP1)by usingahomogenouscell-basedcAMPassay,J.Biomol.Screen.18(2013) 670–677.

[8]S.Titus,S.Neumann,W.Zheng,N.Southall,S.Michael,C.Klumpp,etal., Quantitativehigh-throughputscreeningusingalive-cellcAMPassayidentifies small-moleculeagonistsoftheTSHreceptor,J.Biomol.Screen.13(2008) 120–127.

[9]M.D.Houslay,G.Milligan,TailoringcAMP-signallingresponsesthrough isoformmultiplicity,TrendsBiochem.Sci.22(1997)217–224.

[10]K.B.Seamon,W.Padgett,J.W.Daly,Forskolin:uniquediterpeneactivatorof adenylatecyclaseinmembranesandinintactcells,Proc.Natl.Acad.Sci.U.S.A.

78(1981)3363–3367.

[11]Y.Wang,Y.Kong,G.J.Shei,L.Kang,M.E.Cvijic,Developmentofacyclic adenosinemonophosphateassayforGi-coupledGprotein-coupledreceptors byutilizingtheendogenouscalcitoninactivityinChinesehamsterovarycells, AssayDrugDev.Technol.9(2011)522–531.

[12] S.J.Hill,C.Williams,L.T.May,InsightsintoGPCRpharmacologyfromthe measurementofchangesinintracellularcyclicAMP;advantagesandpitfallsof differingmethodologies,Br.J.Pharmacol.161(2010)1266–1275.

[13]F.Fan,B.F.Binkowski,B.L.Butler,P.F.Stecha,M.K.Lewis,K.V.Wood,Novel geneticallyencodedbiosensorsusingfireflyluciferase,ACSChem.Biol.3 (2008)346–351.

[14]T.Wittenberger,H.C.Schaller,S.Hellebrand,Anexpressedsequencetag(EST) dataminingstrategysucceedinginthediscoveryofnewG-proteincoupled receptors,J.Mol.Biol.307(2001)799–813.

[15]W.He,F.J.Miao,D.C.Lin,R.T.Schwandner,Z.Wang,J.Gao,etal.,Citricacid cycleintermediatesasligandsfororphanG-protein-coupledreceptors,Nature 429(2004)188–193.

[16]I.Toma,J.J.Kang,A.Sipos,S.Vargas,E.Bansal,F.Hanner,etal.,Succinate receptorGPR91providesadirectlinkbetweenhighglucoselevelsandrenin releaseinmurineandrabbitkidney,J.Clin.Invest.118(2008)2526–2534.

[17]T.Rubic,G.Lametschwandtner,S.Jost,S.Hinteregger,J.Kund,N.Carballido- Perrig,etal.,TriggeringthesuccinatereceptorGPR91ondendriticcells enhancesimmunity,Nat.Immunol.9(2008)1261–1269.

[18]P.Sapieha,M.Sirinyan,D.Hamel,K.Zaniolo,J.S.Joyal,J.H.Cho,etal.,The succinatereceptorGPR91inneuronshasamajorroleinretinalangiogenesis, Nat.Med.14(2008)1067–1076.

[19]N.Sadagopan,W.Li,S.L.Roberds,T.Major,G.M.Preston,Y.Yu,etal.,Circulating succinateiselevatedinrodentmodelsofhypertensionandmetabolicdisease, Am.J.Hypertens.20(2007)1209–1215.

[20]P.R.Correa,E.A.Kruglov,M.Thompson,M.F.Leite,J.A.Dranoff,M.H.Nathanson, Succinateisaparacrinesignalforliverdamage,J.Hepatol.47(2007)262–269.

[21] C.Hogberg,O.Gidlof,C.Tan,S.Svensson,J.Nilsson-Ohman,D.Erlinge,etal., SuccinateindependentlystimulatesfullplateletactivationviacAMPand phosphoinositide3-kinase-betasignaling,J.Thromb.Haemost.9(2011) 361–372.

[22]B.Spath,A.Hansen,C.Bokemeyer,F.Langer,Succinatereversesin-vitro plateletinhibitionbyacetylsalicylicacidandP2Yreceptorantagonists, Platelets23(2012)60–68.

[23]D.Bhuniya,D.Umrani,B.Dave,D.Salunke,G.Kukreja,J.Gundu,etal., DiscoveryofapotentandselectivesmallmoleculehGPR91antagonist,Bioorg.

Med.Chem.Lett.21(2011)3596–3602.

[24]H.Takakura,M.Hattori,M.Takeuchi,T.Ozawa,Visualizationandquantitative analysisofGprotein-coupledreceptor-beta-arrestininteractioninsinglecells andspecificorgansoflivingmiceusingsplitluciferasecomplementation,ACS Chem.Biol.7(2012)901–910.

[25]J.Hanson,N.Ferreiros,B.Pirotte,G.Geisslinger,S.Offermanns,Heterologously expressedformylpeptidereceptor2(FPR2/ALX)doesnotrespondtolipoxinA (4),Biochem.Pharmacol.85(2013)1795–1802.

[26]J.H.Zhang,T.D.Chung,K.R.Oldenburg,Asimplestatisticalparameterforusein evaluationandvalidationofhighthroughputscreeningassays,J.Biomol.

Screen.4(1999)67–73.

[27]V.V.Gurevich,E.V.Gurevich,Thestructuralbasisofarrestin-mediated regulationofG-protein-coupledreceptors,Pharmacol.Ther.110(2006) 465–502.

[28]C.Southern,J.M.Cook,Z.Neetoo-Isseljee,D.L.Taylor,C.A.Kettleborough,A.

Merritt,etal.,Screeningbeta-arrestinrecruitmentfortheidentificationof naturalligandsfororphanG-protein-coupledreceptors,J.Biomol.Screen.18 (2013)599–609.

[29]B.K.Atwood,J.Lopez,J.Wager-Miller,K.Mackie,A.Straiker,ExpressionofG protein-coupledreceptorsandrelatedproteinsinHEK293,AtT20,BV2,and N18celllinesasrevealedbymicroarrayanalysis,BMCGenomics12(2011)14.

[30]C.C.Bleul,M.Farzan,H.Choe,C.Parolin,I.Clark-Lewis,J.Sodroski,etal.,The lymphocytechemoattractantSDF-1isaligandforLESTR/fusinandblocksHIV- 1entry,Nature382(1996)829–833.

[31]C.G.Wermuth,Selectiveoptimizationofsideactivities:theSOSAapproach, DrugDiscov.Today11(2006)160–164.

[32]M.T.Klein,M.Dukat,R.A.Glennon,M.Teitler,Towardselectivedrug developmentforthehuman5-hydroxytryptamine1Ereceptor:acomparison of5-hydroxytryptamine1Eand1Freceptorstructure-affinityrelationships,J.

Pharmacol.Exp.Ther.337(2011)860–867.

[33]C.W.Scott,M.F.Peters,Label-freewhole-cellassays:expandingthescopeof GPCRscreening,DrugDiscov.Today15(2010)704–716.

[34]T.P.Kenakin,Cellularassaysasportalstoseven-transmembranereceptor- baseddrugdiscovery,Nat.Rev.DrugDiscov.8(2009)617–626.

[35]S.B.Liggett,N.J.Freedman,D.A.Schwinn,R.J.Lefkowitz,Structuralbasisfor receptorsubtype-specificregulationrevealedbyachimericbeta3/beta2- adrenergicreceptor,Proc.Natl.Acad.Sci.U.S.A.90(1993)3665–3669.

[36]W.Cao,L.M.Luttrell,A.V.Medvedev,K.L.Pierce,K.W.Daniel,T.M.Dixon,etal., Directbindingofactivatedc-Srctothebeta3-adrenergicreceptorisrequired forMAPkinaseactivation,J.Biol.Chem.275(2000)38131–38134.

[37]G.E.Callander,W.G.Thomas,R.A.Bathgate,ProlongedRXFP1andRXFP2 signalingcanbeexplainedbypoorinternalizationandalackofbeta-arrestin recruitment,Am.J.Physiol.CellPhysiol.296(2009)C1058–66.

[38]S.Al-Sabah,M.Al-Fulaij,G.Shaaban,H.A.Ahmed,R.J.Mann,D.Donnelly,etal., TheGIPreceptordisplayshigherbasalactivitythantheGLP-1receptorbut doesnotrecruitGRK2orarrestin3effectively,PLoSOne9(2014)e106890.

[39]S.Rajagopal,J.Kim,S.Ahn,S.Craig,C.M.Lam,N.P.Gerard,etal.,Beta-arrestin- butnotGprotein-mediatedsignalingbythedecoyreceptorCXCR7,Proc.Natl.

Acad.Sci.U.S.A.107(2010)628–632.

[40]W.deLau,N.Barker,T.Y.Low,B.K.Koo,V.S.Li,H.Teunissen,etal.,Lgr5 homologuesassociatewithWntreceptorsandmediateR-spondinsignalling, Nature476(2011)293–297.

[41]L.Sundstrom,P.J.Greasley,S.Engberg,M.Wallander,E.Ryberg,Succinate receptorGPR91,aGalpha(i)coupledreceptorthatincreasesintracellular calciumconcentrationsthroughPLCbeta,FEBSLett.587(2013)2399–2404.

[42]C.Harrison,J.R.Traynor,The[35S]GTPgammaSbindingassay:approachesand applicationsinpharmacology,LifeSci.74(2003)489–508.

[43]E.Kostenis,M.Waelbroeck,G.Milligan,Techniques:promiscuousGalpha proteinsinbasicresearchanddrugdiscovery,TrendsPharmacol.Sci.26(2005) 595–602.

[44]J.Pantel,S.Y.Williams,D.Mi,J.Sebag,J.D.Corbin,C.D.Weaver,etal., Developmentofahighthroughputscreenforallostericmodulatorsof melanocortin-4receptorsignalingusingarealtimecAMPassay,Eur.J.

Pharmacol.660(2011)139–147.

[45]R.Irannejad,J.C.Tomshine,J.R.Tomshine,M.Chevalier,J.P.Mahoney,J.

Steyaert,etal.,ConformationalbiosensorsrevealGPCRsignallingfrom endosomes,Nature495(2013)534–538.

[46]B.F.Binkowski,B.L.Butler,P.F.Stecha,C.T.Eggers,P.Otto,K.Zimmerman,etal., AluminescentbiosensorwithincreaseddynamicrangeforintracellularcAMP, ACSChem.Biol.6(2011)1193–1197.

[47]S.P.Brothers,S.A.Saldanha,T.P.Spicer,M.Cameron,B.A.Mercer,P.Chase,etal., Selectiveandbrainpenetrantneuropeptideyy2receptorantagonists discoveredbywhole-cellhigh-throughputscreening,Mol.Pharmacol.77 (2010)46–57.

[48]P.A.Insel,R.S.Ostrom,Forskolinasatoolforexaminingadenylylcyclase expression,regulation,andGproteinsignaling,Cell.Mol.Neurobiol.23(2003) 305–314.

[49]J.R.Jasper,M.C.Michel,P.A.Insel,AmplificationofcyclicAMPgeneration revealsagonisticeffectsofcertainbeta-adrenergicantagonists,Mol.

Pharmacol.37(1990)44–49.