Vrije Universiteit Brussel

Internalized Receptor for Glucose-dependent Insulinotropic Peptide stimulates adenylylcyclase on early endosomesIsmail, Sadek; Gherardi, Marie-Julie; Froese, Alexander; Zanoun, Madjid; Gigoux, Véronique;Clerc, Pascal; Gaits-Iacovoni, Frederique; Steyaert, Jan; Nikolaev, Viacheslav O; Fourmy,DanielPublished in:Biochemical Pharmacology

DOI:10.1016/j.bcp.2016.09.009

Publication date:2016

License:Unspecified

Document Version:Final published version

Link to publication

Citation for published version (APA):Ismail, S., Gherardi, M-J., Froese, A., Zanoun, M., Gigoux, V., Clerc, P., ... Fourmy, D. (2016). InternalizedReceptor for Glucose-dependent Insulinotropic Peptide stimulates adenylyl cyclase on early endosomes.Biochemical Pharmacology, 120, 33-45. https://doi.org/10.1016/j.bcp.2016.09.009

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portalTake down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Download date: 10. Aug. 2021

Biochemical Pharmacology 120 (2016) 33–45

Contents lists available at ScienceDirect

Biochemical Pharmacology

journal homepage: www.elsevier .com/locate /b iochempharm

Internalized Receptor for Glucose-dependent Insulinotropic Peptidestimulates adenylyl cyclase on early endosomes

http://dx.doi.org/10.1016/j.bcp.2016.09.0090006-2952/� 2016 Elsevier Inc. All rights reserved.

Abbreviations: GIPR, glucose-dependent Insulinotropic Peptide Receptor; cAMP,30 ,50-cyclic adenosine monophosphate; BRET, Bioluminescence Resonance EnergyTransfer; FRET, Fluorescence Resonance Energy Transfer.⇑ Corresponding author at: Laboratoire de Physique et Chimie des Nano-objets

(LPCNO), team RTTC, Université de Toulouse, CNRS, INSA, INSERM, Université PaulSabatier, 1 avenue Jean Poulhès, 31432 Toulouse Cedex, France.

E-mail address: Daniel.Fourmy@inserm.fr (D. Fourmy).URL: http://www.rttc-research.com (D. Fourmy).

Sadek Ismail a, Marie-Julie Gherardi a, Alexander Froese b, Madjid Zanoun c, Véronique Gigoux a,Pascal Clerc a, Frederique Gaits-Iacovoni d, Jan Steyaert e,f, Viacheslav O. Nikolaev b, Daniel Fourmy a,⇑a Laboratoire de Physique et Chimie des Nano-objets (LPCNO), team RTTC, Université de Toulouse, CNRS, INSA, INSERM, Université Paul Sabatier, Toulouse, FrancebGerman Center for Cardiovascular Research, University Medical Center Hamburg-Eppendorf and Institute of Experimental Cardiovascular Research, Hamburg, GermanycCellular Imaging Facility Rangueil, INSERM U1048/I2MC, Toulouse, Franced INSERM, UMR1048, University of Toulouse 3, Institute of metabolic and cardiovascular diseases, Toulouse, Francee Structural Biology Brussels, Vrije Universiteit Brussel, Brussel, Belgiumf Structural Biology Research Center, VIB, Brussel, Belgium

a r t i c l e i n f o a b s t r a c t

Article history:Received 20 July 2016Accepted 12 September 2016Available online 15 September 2016

Keywords:Glucose-dependent-insulinotropic receptor,InternalizationcAMPBRETFRETEarly endosomes

Until very recently, G-protein dependent signal of GPCRs was thought to originate exclusively from theplasma membrane and internalized GPCRs were considered silent. Here, we demonstrated that, onceinternalized and located in the membrane of early endosomes, glucose-dependent Insulinotropic recep-tor (GIPR) continues to trigger production of cAMP and PKA activation. Direct evidence is based on iden-tification of the active form of Gas in early endosomes containing GIPR using a genetically encoded GFPtagged nanobody, and on detection of a distinct FRET signal accounting for cAMP production at the sur-face of endosomes containing GIP, compared to endosomes without GIP. Furthermore, decrease of thesustained phase of cAMP production and PKA activation kinetics as well as reversibility of cAMP produc-tion and PKA activity following GIP washout in cells treated with a pharmacological inhibitor of GIPRinternalization, and continuous increase of cAMP level over time in the presence of dominant-negativeRab7, which causes accumulation of early endosomes in cells, were noticed. Hence the GIPR joins thefew GPCRs which signal through G-proteins both at plasma membrane and on endosomes.

� 2016 Elsevier Inc. All rights reserved.

1. Introduction (30,50-cyclic adenosine monophosphate, cAMP or inositol triphos-

Seven-transmembrane receptors, also termed G-protein-coupled receptors (7 TMRs or GPCRs), form the largest class of cellsurface membrane receptors, involving several hundred membersin the human genome. Biological effects triggered by GPCRs resultfrom activation of both G-protein-dependent and G-protein-independent intracellular signaling pathways [1–3]. Until veryrecently, production of diffusible second messengers was thoughtto exclusively originate from the cell plasma membrane, whereembedded GPCRs bound their cognate ligands and are subse-quently activated to stimulate production of second messengers

phate, IP3) and downstream signaling cascade [3]. On the otherhand, activation of membrane GPCRs is generally immediately fol-lowed by their Ser/Thr phosphorylation by second messenger-dependent kinases and G-protein Receptor Kinases (termed GRKs)[1–4]. Phosphorylated GPCRs then recruit adaptor proteins such asb-arrestins which uncouple G-proteins from membrane receptorsand promote GPCR endocytosis. Alternatively, activated and phos-phorylated GPCRs can interact directly with components of theendocytic machinery which generally involves clathrin and the cla-thrin adaptor AP2. Coated-pits containing GPCRs are then sepa-rated from the plasma membrane by action of dynamin, leadingto the formation of early endosomes. Finally, internalized GPCRsare sorted via the endocytic pathway to lysosomes or are recycledto the cell surface [1–4]. Thus, consensus holds that internalizationdisrupts G-protein-dependent signaling of GPCRs and conse-quently, internalized GPCRs become silent with respect toG-protein dependent production of diffusible second messengers.

Recently, the classical concept whereby G-protein-dependentsignaling of GPCRs is restricted to plasma membrane has been

34 S. Ismail et al. / Biochemical Pharmacology 120 (2016) 33–45

challenged and refuted. Evidence of endosomal G-protein signalingof GPCRs was obtained with the parathyroid hormone receptor(PTHR) which induces sustained or persistent production of cAMPin response to PTH stimulation [5]. In kinetic studies, the sustainedphase of cAMP production persisted after agonist removal. Con-comitantly, sustained production of cAMP, ascribed to internalizedThyroid Stimulating Hormone Receptor (TSHR), was documentedin native thyroid follicles isolated from transgenic mice expressinga FRET sensor of cAMP [6]. Dopamine 1 receptor and vasopressinreceptors were subsequently shown to trigger cAMP productionfrom endosomes [7,8]. Among the different experimental argu-ments provided in support of the concept of endosomal cAMP pro-duction, chemical and genetically encoded inhibitors of GPCRinternalization such as dynasore and dynamin dominant-negativemutants, were shown to affect the sustained phase of cAMP pro-duction [5,6]. Additional and more direct evidences were providedin studies with the b2-adrenergic receptor in which conformationalbiosensors, consisting of nanobodies, enabled identification, byconfocal microscopy, of the active form of the b2-receptor, as wellas the active Gas subunit in early endosomes of living cells [9].Interestingly, in agreement with the concept of compartmentalizedcAMP signaling originally proposed in a study on cardiac myocytes,accumulating data support that cAMP produced from endosomesby activated GPCRs causes qualitatively different physiologicaleffects compared to cAMP generated from the plasma membrane[5,6,10–12].

However, despite reasonably convincing evidence that internal-ized GPCRs continue to trigger G-protein dependent signals, thisview has been refuted and still remains a subject of controversy[13,14]. Furthermore, G-protein-dependent production of dif-fusible second messengers by GPCRs in endosomes was onlyreported for a very restricted number of GPCRs (recently reviewedin [15]). Additionally, direct detection of cAMP production onendosomes resulting from activation of an internalized GPCR wasnot provided so far. This prompted us to investigate if theGlucose-dependent Insulinotropic Receptor (GIPR) which is aphysiologically and pharmacologically important receptor regulat-ing glucose and lipid homeostasis [16] and which is a universalGPCR overexpressed in neuroendocrine tumors [17], would alsobe endowed of capability to signal from early endosomes.

GIPR which belongs to the subfamily-2 of GPCR, triggers Gs-mediated cAMP production and subsequent signaling cascades[16]. We have shown that GIPR undergoes rapid abundant inter-nalization following stimulation by GIP and that internalized GIPRis mainly sorted to the lysosomal degradation pathway [18]. Inter-estingly, in contrast to most GPCRs, including the closely struc-turally and functionally related GLP1 receptor, GIPR isinternalized independently of b-arrestin recruitment [18,19].

In the current study, by using an integrative approach com-prised of live cell confocal microscopy, pharmacological and genet-ically encoded tools, BRET and FRET measurement of cAMP andPKA activity in whole cells, immune-detection of activated formof Gs and FRET cAMP measurements at the endosome surface(see Fig. 1), we unequivocally demonstrate that internalized GIPRremains active and triggers cAMP production at endosomes. Thus,GIPR joins the group of GPCRs which are activated at the plasmamembrane and once, activated and internalized continue to stimu-late production of their cognate diffusible second messenger.

2. Materials and methods

2.1. Materials

Fragment 1–30 of human GIP (termed GIP) and DY647 labeled-GIP (termed DY647-GIP) were obtained as previously described

[20]. The fluorescent probe was highly specific of GIPR (less than5% nonspecific labeling in the present of 100-fold excess of unla-beled peptide). Radio-labeled GIP was obtained by radio-iodination of Phe1-GIP(1–30) with 125I-Na (Perkin Elmer, France)in the presence of chloramine T and was HPLC purified on a C-18column. 125I-Phe1-GIP bound to a single class of GIPR binding sitesfrom HEK 293T or Flp-InTMGIPR-293 cells with a dissociation con-stant, Kd of 75.7 ± 8.4 nM. Sequence encoding short variant of thehuman GIPR was derived from a plasmid kindly given by ProfessorBernard Thorens (Lausanne, Switzerland) [21]. Chemicals werefrom the following sources: Dyngo-4a from Abcam (Cambridge,UK), Forskolin from Sigma-Aldrich (St. Quentin Fallavier, France).

Plasmids encoding DsRed tagged Rab5 (termed DsRed-Rab5),DsRed tagged Rab7-DN (termed DsRed-Rab7-DN), GFP taggedEEA1 (termed GFP-EEA1) were supplied by Addgene (www.ad-dgene.org). Plasmid for BRET measurements of cAMP production,namely Rluc-Epac1-citrine was kindly provided by Professor MarcCaron.

The cytosolic cAMP sensor Epac1-camps sequence [22] wasused as a backbone to construct the EYFP-EPAC1-ECFP-FYVE sensortargeted to early endosomes. After removal of stop codon, the FYVEsequence was amplified by PCR using the primers AAG GAT CCATGC CCT TGG TGG ATT TCT TCT GCT GGC AAT CTA GTC AAC GGand aaa ctc gag ttatccttgcaagtcattgaaaca, and inserted via BamHIand XhoI restriction sites into the Epac1-camps vector backbone.The resultant targeted sensor contained the full Epac1-campssequences followed by a flexible linker GSMPLVDFFC and the FYVEsequence from WQSSQ on. Plasmid to detect PKA activity encodedA Kinase Activity Reporter (AKAR3) sensor. It was constructed onthe basis of previous report [23]. Plasmid encoding the nanobodyNb-37 recognizing activated form of Gs was generated accordingto [24]. Nb-37 specifically recognizes nucleotide-free form of asubunit of Gs [9].

2.2. Cell lines and transfections

HEK 293 cells stably expressing the GIPR (Flp-lnTM HEK-GIPR)were obtained using the Flp-InTM system (Invitrogen). The cell lineswere maintained in Dulbecco’s Modified Eagle’s medium supple-mented with 10% of fetal bovine serum (FBS), in a humidifiedatmosphere at 95% air and 5% CO2. Transfections for BRET experi-ments were performed using polyethylenimine (PEI) transfectionreagent (1 mg/mL, pH 7.4) (Polyplus, Illkirch, France). Plasmidswere diluted in DMEM without FBS (ratio DNA (lg)/PEI (lL) 1:3).The mixture was mixed for 15 s on a vortex, incubated for15 min at room temperature and then deposited on the cells. Forconfocal experiments, transfections were performed using Lipofec-tamin 2000 (Invitrogen Life Technologies) following provider’sinstructions (ratio DNA (lg)/LPF2000 (lL) 1:2).

2.3. BRET assay of cAMP production

Flp-InTMGIPR-293 cells were plated onto 10-cm culture dishesand overnight grown afterward they were co-transfected with atotal amount of 5 lg DNA plasmid comprising 4 lg of cAMP BRETbiosensor RLuc-Epac-Citrine completed with 1 lg of non-codingplasmid. 24 h after transfection, cells were plated in 96-well clearbottom plates (Corning) at a density of 100.000 cells per well inphenol red free DMEM 2% FBS. After an overnight incubation, themedium was removed and replaced by calcium and magnesiumfree PBS. BRET assay was initiated by adding 10 ll of coelenter-azine h to the wells (final concentration 5 lM). After 10 min ofincubation with coelenterazine h, stimulant of cAMP production,namely GIP or Forskolin was injected. Live-time measurementswere recorded at 37 �C every 30 s for 60 min. Luminescence andfluorescence readings were performed on a Mithras LB940 instru-

4. Dyngo4a

Clathrin- coated

pits

Early endosomes

cAMP

Rab-5+ EEA1+

AC cAMP

Plasma membrane

AP2

AP2

Dynamin GIP/GIPR

RLuc

a

2. Cytosolic cAMP Bret or Fret sensor

a

Y F P 7. Endosomal

cAMP Fret sensor eCFP-FYVE

1. DY647-GIP

6. Nanobody

5. DN-Rab-7

G*s

PI(3)P

GFP-Nb-37

3. DsRed-Rab-5 GFP-EEA1 C

a

Late endosomes/ lysosomes

PKA

8. PKA FRET sensor

e

eCFP

eYFP eCFP

P

Fig. 1. Schematic representation of GIP receptor internalization and post-endocytic trafficking and tools used to identify endosomal production of cAMP and PKA activity.GIP-bound-GIPR was identified using DY647-GIP (1). Total cAMP production was measured using cytosolic Bret sensor RLuc-Epac1-citrine or Fret sensor EYFP-Epac1-ECFP (2).Early endosomes were labeled with DsRed-Rab5 or GFP-EEA1 (3). Dyngo4-a (4), an inhibitor of dynamin, served to inhibit GIPR internalization whereas DsRed-DN-Rab7 (5)caused accumulation of early endosomes. Activity of GIPR in early endosomes was detected using a nanobody specific of active form of Gas subunit (6) and an endosomalFRET sensor genetically targeted to early endosomes thanks to FYVE sequence recognizing PI(3)P (7). Activity of PKA was detected using AKAR3 FRET sensor (8).

S. Ismail et al. / Biochemical Pharmacology 120 (2016) 33–45 35

ment (Berthold France, Thoiry, France) that allows the sequentialintegration of signals at 465–505 nm and 515–555 nm windows.MicroWin 2000 software was used for calculation.

2.4. Live cell confocal fluorescence imaging

After an overnight transfection with 2 lg/well of pcDNA5/FRTcontaining cDNAs of interest, cells were transferred onto poly-L-lysine (Sigma-Aldrich) coated 4-well Lab-Tek chambered cover-glass (Nunc). 24 h later, the culture medium was replaced by phe-nol red free medium (DMEM1X, 4.5 g/L D-glucose, L-Glutamine,25 mM HEPES, without sodium pyruvate, gibco Lifetechnologies).Cells were stimulated with appropriate ligand and were imagedat 37 �C with a confocal Zeiss Laser Scanning Microscope LSM-780 equipped with 63�/1.4 NA oil immersion objective, a GaAsPdetector with quantum yield of 45% and a parallel spectral detec-tion attending 32 channels simultaneously in lambda (k) mode.For multi-color imaging, the fluorophores were excited by the cor-responding lasers and the fluorescence signals were collected inthe corresponding emission spectra simultaneously. Images wereprocessed using ImageJ software.

2.5. Confocal microscopy image quantification of colocalisation

For co-localization determination both visual inspection andJACOP Image J plugin (Mander’s coefficient) were used. Briefly,for evaluating the co-localization of DY647-GIP with DsRed-Rab5or GFP-EEA1, clearly observed DY647-GIP-containing vesicles wereselected and counted. After merging the DY647 channel withDsRed or GFP channel using Image J, vesicles showing co-localization were identified and counted. At the different timesafter stimulation, co-localization level representing co-localizingvesicles expressed as percent of total DY647-GIP-containing endo-cytic vesicles was determined. We found that Mander’s coefficientsdetermined by JACOP ImageJ plugin matched well with the co-localization levels determined by visual inspection. 3–5 cells wereanalyzed in 3–4 independent experiments.

2.6. FRET assays of cAMP production and PKA activity

cAMP production was determined by FRET on Flp-InTMGIPR-293 cells transfected with cytosolic cAMP sensor Epac1-campsplasmid [22]. FRET was monitored as the ratio between emission

36 S. Ismail et al. / Biochemical Pharmacology 120 (2016) 33–45

at 535 ± 20 nm (YFP) and emission at 480 ± 615 nm (CFP), uponexcitation at 436 ± 10 nm using MetaFluor 5.0 software (MolecularDevices). The imaging data were analyzed utilizing MetaMorph 5.0(Molecular Devices) and Prism (GraphPad Software) software, bycorrecting for spillover of CFP into the 535-nm channel and directYFP-excitation, to give corrected YFP/CFP ratio data. Images wereacquired every 5 s, with 5-ms illumination time, which resultedin negligible photobleaching for over 30-min observation. To studyGIP-induced changes and reversibility in cAMP, cells were contin-uously superfused with phenol red–free medium containing 1%BSA or the same plus GIP and/or internalization inhibitor, Dyngo-4a, with a custom apparatus. All experiments were performed at25 �C.

FRET assay of cAMP from endosomes was carried out on a con-focal Zeiss Laser Scanning Microscope LSM-780. FRET was moni-tored as the ratio between emission at 539 ± 23 nm (YFP) andemission at 482 ± 18 nm (CFP) upon excitation at 458 nm. TheDY647-GIP emission at 655 ± 40 nm was detected in sequentialmode upon excitation at 633 nm. By using this setup and underour experiment conditions, we verified experimentally that GIP-DY647 fluorochrome did not influence signal of cAMP FRET biosen-sor (not shown). Images were analyzed using ImageJ. Thresholdlimited endosomes were outlined using ‘‘analyze particle” tool.Mean FRET value for each endosome is calculated on the FRETresulting image. Endosome state, namely internalized (containingDY647-GIP) or pre-existing (free of ligand), was visually deter-mined on the merged image.

PKA activity was detected using FRET-based A Kinase ActivityReporter (AKAR3) sensor previously introduced [23]. This sensorwas composed of a pair of fluorescent moieties (eCFP and eYFP)sandwiching a molecular probe composed of a surrogate substratefor PKA and a phosphoamino acid binding domain (FHA). Followingphosphorylation of the substrate by PKA, the FHA domain binds thephophorylated amino acid causing a conformational change whichresults in FRET signal increase. FRET was monitored and analyzedas described for cAMP measurements.

2.7. Receptor binding assays

Flp-InTMGIPR-293 cells were grown onto 10-cm culture dishesfor 24 h and then transferred to 24-well plates. Approximately24 h later, binding assays were performed using 125I-Phe1-GIPaccording to the protocol previously described in detail [25]. Nonspecific binding corresponded to residual binding of 125I-Phe1-GIPin the presence of 1 lM GIP. IC50 (concentration inhibiting half ofspecific binding) was calculated using the non-linear curve fittingsoftware GraphPad Prism (San Diego, CA).

2.8. Statistics

All values are expressed as the mean ± standard error ofthe mean (SEM). Statistical analyses of data using Student’s t-testwere performed with GraphPad Prism software version 6.0.Significance degrees were given as following: *0.01 < p < 0.05;**0.001 < p < 0.01; ***p < 0.001.

3. Results

3.1. Inhibition of GIPR sorting to early endosomes affected thesustained phase of cAMP production

We investigated production of cAMP from early endosomes inHEK cells which has been proven to be an excellent and relevantcell model for the study of GPCR internalization and related signal-ing. HEK 293 cells stably expressing GIPR, (termed Flp-lnTM HEK-

GIPR), bound GIP with a nanomolar affinity and dose-dependently responded to GIP by stimulating production of cAMPwith an EC50 of 1.1 ± 0.2 nM [18]. We traced GIPR at the cell surfaceor interior by using fluorescently tagged GIP (termed DY647-GIP)which bound to GIPR and stimulated cAMP production identicallyto unmodified GIP and, moreover, co-localized with GIPR for up to2 h during post-endocytic sorting to lysosomes [18]. As shown onFig. 2A, confocal microscopy observations of Flp-lnTM HEK-GIPRcells revealed intense homogenous labeling of the plasma mem-brane immediately after addition of DY647-GIP. Fluorescencewas then rapidly relocated in membrane clusters and rapidlyentered into the cells as punctuate vesicles which moved overthe incubation time. Confocal microscopy imaging of cells express-ing tagged functional proteins of early endosomes, namely DsRed-Rab5 or GFP-EEA1, indicated that chronic stimulation of GIPRresulted in abundant localization of DY647-GIP-bound GIPR invesicles positive for the markers of early endosomes (Fig. 2B–D).

In order to determine whether GIPR can signal from endosomesto produce cAMP, we first sought to assess the influence of inter-nalization blockade on the kinetic and level of GIP-induced cAMPresponse measured in real-time by a cytosolic BRET biosensor.Internalization of GIPR occurs through clathrin-coated pits, withan involvement of dynamin that enables detachment of endocyticvesicles from the plasma membrane and formation of early endo-somes containing GIPR [18] (Fig. 1). In light of previous data show-ing that the chemical dynamin inhibitor Dynasore was not fullyeffective to arrest GIPR internalization, we tested Dyngo-4a whichwas reported to be �6-fold more potent than dynasore to inhibitdynamin in intact cells [26].

We therefore determined if Dyngo-4a affected kinetic and real-time levels of GIP-stimulated cAMP formation. As shown on Fig. 3,cAMP stimulated by 10 nM GIP rapidly increased and remainedsustained at its maximum level over time. In contrast, cAMP levelcontinuously declined in the presence of Dyngo-4a to a value of�27% of the maximum value at 60 min of stimulation, suggestingthe existence of a relationship between cAMP production and GIPRinternalization. Profile of cAMP production stimulated by 100nMGIP was slightly different. It showed a rapid increase followed bya slight drop, likely due to strong desensitization of the initialcAMP response at this GIP concentration until 10 min of stimula-tion [18]. Then, cAMP level remained constant to �70% of the max-imum value.

We also assessed that sustained phase of cAMP production trulyresult from internalized GIPR by studying reversibility of cAMPresponse after GIP washout from the cells. For this purpose, cAMPwas measured by FRET using Epac1-Camps cytosolic sensor [22] oncells which were stimulated with 10 nM GIP and then superfusedwith buffer to wash out GIP from the cells. Stimulation by GIPfor 2 or 5 min caused rapid decreases of FRET ratio accountingfor cAMP production and GIP washout did not result in significantFRET changes indicating irreversibility of cAMP response (Fig. 4A).In contrast, similar FRET measurements carried out on cells treatedwith Dyngo-4a to block GIPR internalization, revealed that afterthe decrease of FRET signal in response to GIP stimulation, GIPwashout resulted in an important return of FRET signal accountingfor reversibility of cAMP response (Fig. 4B). Quantitative analysis ofthe data indicates that cAMP response was reversible in cells hav-ing GIPR internalization blocked whereas cAMP response was onlyweakly reversible in control cells internalizing GIPR (Fig. 4C).

To further assess sustained GIP-induced cAMP production, wemeasured activity of protein kinase A (PKA), a key effector forcAMP signaling, using AKAR3, a FRET sensor of PKA activity. Asillustrated on Fig. 5A, stimulation by GIP for 2 min caused rapidincrease of FRET, and GIP washout did not result in FRET decrease.This indicates that PKA was rapidly activated by GIP and kinaseactivity remained stable after GIP washout. In contrast, on cells

DY647-GIP

DsRed-Rab5

DY647-GIP/ DsRed-Rab5

* PCC = 0.74 0.04

GFP-EEA1

0’ 20’

’03’3 20’5’0’

DY647-GIP

DY647-GIP/ GFP-EEA1

A

B C

0.6 0.8 1.0 1.2 1.4

20

40

60

80

100

Position (μm)

D

Rel

ativ

e flu

ores

cenc

e (%

max

)

Fig. 2. Internalization and localization of GIP receptor in early endosomes. HEK cells expressing the human GIP receptor (Flp-InTMGIPR-293 cells) transfected with plasmidsencoding fluorescence tagged proteins of the early endosomes, DsRed-Rab5 or GFP-EEA1, were stimulated with 10 nM DY647-GIP at 37 �C. Fluorescence confocal microscopyimages were acquired over the time using identical instrument settings. Panel A corresponds to a representative microscopy field showing two cells, over-expressing or notDsRed-Rab5 (displayed in green). DY647-GIP-labeled GIPR (displayed in red) is seen at the cell surface at time 3 min of incubation, then, in endocytic vesicles close to themembrane (5 min) and scattered within the cells (20, 30 min). DY647-GIP-labeled GIPR co-localizes with DsRed-Rab5 (79 ± 9% at 30 min, Pearson’s correlation coefficient -PCC- is indicated on image of time 30 min). Panel B shows co-localization of DY647-GIP-labeled GIPR with GFP-EEA1 at times 20 and 60 min of stimulation (values of co-localization were 73 ± 17%, PCC: 0.64 ± 0.07 and 24 ± 4%, PCC: 0.50 ± 0.06, respectively). Panel C shows a zoom of an endosome containing both DY647-GIP and DsRed-Rab5,and panel D shows its line scan. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

S. Ismail et al. / Biochemical Pharmacology 120 (2016) 33–45 37

treated with Dyngo-4a, FRET signal in response to GIP stimulationreturned to basal value after GIP washout (Fig. 5A). Hence, PKAactivity resulting from transient stimulation of GIPR, presumablyat the plasma membrane, was reversible. Quantitative analysis ofthe data confirms that PKA activity was reversible in cells havingGIPR internalization blocked whereas it was only weakly reversiblein cells internalizing GIPR (Fig. 5B).

Control experiments were carried out in order to check inhibi-tory action of Dyngo4-a on GIPR internalization. Confocal micro-scopy imaging experiments showed a dramatic decrease of Flp-lnTM HEK-GIPR cell membrane labeling by DY647-GIP in the pres-ence of Dyngo-4a (not shown). This observation, which could bedue to quenching of DY647-fluorescence in the presence ofDyngo4-a precluded the use of DY647-GIP to assess inhibitory

0 20 40 60

50

100

Stimulation time (min)

cAM

P r

espo

nse

(% M

ax)

(1/B

RE

T)

Control

+ Dyngo-4a + 100 nM GIP

0 20 40 60

50

100Control

+ Dyngo-4a

+ 10 nM GIP

Fig. 3. Inhibition of GIP receptor internalization affected the sustained phase ofGIP-stimulated cAMP production. HEK cells expressing the human GIP receptor(Flp-InTMGIPR-293 cells) transfected with plasmids encoding BRET biosensor, Rluc-Epac1-citrine, were stimulated with GIP (10 or 100 nM) for BRET measurement ofcAMP, as described in ‘‘Materials and Methods” section. Blue records (control)correspond to 1/BRET signal from experiments performed in the absence ofinhibitor of internalization, and Red records correspond to 1/BRET signals in thepresence of inhibitor of internalization (Dyngo4-a, 60 lM). 1/BRET values expressedas percent of maximum in control assay are the mean ± SEM of 3 individualexperiments. They account for a decrease of the sustained and late phase of cAMPproduction in the presence of Dyngo-4a. (For interpretation of the references tocolor in this figure legend, the reader is referred to the web version of this article.)

38 S. Ismail et al. / Biochemical Pharmacology 120 (2016) 33–45

effect of Dyngo-4a on GIPR internalization. Therefore, cells weretransfected to express GFP-GIPR and then, were stimulated with1 lM GIP in the presence or absence of Dyngo-4a. Confocal micro-scopy imaging showed full blocking of GFP-GIPR internalization byDyngo-4a (Fig. 6A). We then considered the possibility that loss ofcell labeling by DY647-GIP in the presence of Dyngo-4a could bedue to an inhibition of GIP binding to GIPR thus explaining cAMPdecline over time. We performed binding of radio-iodinated GIPin the presence or absence of Dyngo-4a. Results demonstrated thatbinding kinetic of radio-iodinated GIP and binding affinity of GIPfor the GIPR were not affected in the presence of Dyngo-4a, clearlyruling out a possibility of interference of the dynamin inhibitorwith GIP binding to GIPR (Fig. 6B, C). Thus, Dyngo-4a most likelychemically reacts with DY647 and therefore quenched DY647 flu-orescence. Finally, we verified that Dyngo4-a did not affect adeny-lyl cyclase activity by stimulating cells by Forskolin in the presenceof Dyngo-4a at concentration of 30 or 60 lM (Fig. 6D, E).

Collectively, these results and controls strongly support theview that internalized GIPR contributes to the sustained phaseand irreversibility of cAMP and PKA responses in Flp-lnTM HEK-GIPR cells.

3.2. Accumulation of early endosomes containing GIPR enhanced thesustained phase of cAMP production

In order to further establish the relationship between the pres-ence of GIPR in early endosomes and cAMP production profiles, weover-expressed a dominant negative of Rab7 in Flp-lnTM HEK-GIPR.Rab7 is a GTPase that, in addition to its regulatory role in cargotransport from late endosomes to lysosomes, is required for earlysorting events (Fig. 1). Dominant negative of Rab7 (DN-Rab7), byinhibiting exchange between Rab5 and Rab7 in early endosomes

leads to accumulation of early endosomes in cells [27]. Accord-ingly, stimulation of cells over-expressing DsRed-DN-Rab7 byDY647-GIP resulted in strong accumulation of DY647-GIP-boundGIPR in endocytic vesicles positive for GFP-EEA1 (Fig. 7A, B). As aquantitative proof of early endosome accumulation, image analysisand quantification indicated that 88 ± 3% of endocytic vesicles con-taining DY647-GIP-bound GIPR were positive for EEA1 in cellsover-expressing DsRed-DN-Rab7 after 30 min of stimulation. Thisco-localization level was 73 ± 16% in cells not expressing DsRed-DN-Rab7. After 60 min of stimulation, 81 ± 3% of endocytic vesiclescontaining DY647-GIP-bound GIPR were positive for EEA1 in cellsover-expressing DsRed-DN-Rab7 whereas this co-localization leveldropped to 24 ± 4% in cells not expressing DsRed-DN-Rab7(Fig. 7C). These data were confirmed by determination of co-localization using Mander’s coefficient (not shown). We then char-acterized the kinetic of cAMP production in Flp-lnTM HEK-GIPR cellstransfected with plasmid encoding DsRed-DN-Rab7 and which,according to fluorescence confocal microcopy observations, over-expressed the encoded protein. As shown on Fig. 7D, after a rapidrise, level of cAMP was stabilized (for 1 or 10 nM GIP stimulation)or slightly declined during few minutes (for 100 nM GIP stimula-tion) and then increased continuously. These profiles of enhancedcAMP production in cells over-expressing DsRed-DN-Rab7 wereobserved for both physiological and pharmacological concentra-tions of GIP, namely from 1 nM to 100 nM GIP. Control experi-ments showed no significant modification of forskolin-stimulatedcAMP levels in cells over-expressing DsRed-DN-Rab7 (Fig. 7E).Therefore, an accumulation of DY647-GIP-bound GIPR in earlyendosomes correlated with a continuous increase of cAMP levelin Flp-lnTM HEK-GIPR cells.

3.3. Identification of the active form of Gas in early endosomescontaining activated GIPR

At this step of the work, the view that internalized GIPR contin-ues to stimulate cAMP production in early endosomes is based ondata showing modulation of kinetic and level of cAMP by Dyngo-4aand Rab7 dominant negative acting respectively on internalizationand accumulation of GIPR in early endosomes. Therefore, our nextaim was to provide more direct evidence by detecting GIPR activa-tion and signaling in early endosomes. We first used a conforma-tional biosensor (termed GFP-Nb37) composed of a single-domain antibody (nanobody) tagged with GFP and recognizingGs only in its active form. In the absence of GIPR stimulation,genetically expressed GFP-Nb37 localized exclusively in the cyto-plasm of cells (Fig. 8A). Immediately after stimulation by GIP,GFP-Nb37 was recruited to the plasma membrane. This recruit-ment of GFP-Nb37 to the plasma membrane lasted during thechronic stimulation by GIP (Fig. 8A). Strikingly, many endocyticvesicles containing DY647-GIP were also labeled by GFP-Nb37,demonstrating presence of active Gas together with DY647-GIP-bound GIPR in these vesicles (Fig. 8A, lower images). Furthermore,vesicles positive for GFP-Nb37 also expressed DsRed-Rab5 or GFP-EEA1, two markers of early endosomes. This co-localization wasillustrated by line scan analysis of confocal microscopy images ofan endocytic vesicle (Fig. 8C). Quantification of co-localizationindicated that 76.0 ± 7.0% of endocytic vesicles containingDY647-GIP-bound GIPR were labeled with GFP-Nb37 at time30 min. This proportion decreased to 48.9 ± 7.9% at 60 min. In cellsover-expressing DsRed-DN-Rab7, the proportion of endocytic vesi-cles labeled with Nb-37-GFP remained high, 82.8 ± 4.1% and78.5 ± 4.7% at times 30 and 60 min, respectively (Fig. 8D, E). Fromthese different experiments, it appears that active Gas is presenttogether with activated GIPR in early endocytic vesicles andover-expression of Rab7 dominant negative prolongs this co-localization.

A, control

B, Dyngo-4a

C, reversibility

Fig. 4. Inhibition of GIPR internalization abolished irreversibility of GIP-induced cAMP production. Stable Flp-InTMGIPR-293 cells were transfected with plasmid encodingcytosolic biosensor Epac1-camps. 24 h post-transfection, cells were washed and then incubated with medium (A, control) or with medium containing 30 lM Dyngo-4a (B,Dyngo-4a). Next, cells were superfused with 10nM GIP for 2 or 5 min with subsequent washout. All experiments were performed at 25 �C. During these experiments, FRETwas recorded as described in ‘‘Materials and Methods” section. Four representative records are shown. Values of YFP/CFP FRET ratios are expressed relative to initial FRETratio which was normalized to 1.00. For analysis of reversibility of cAMP response after GIP washout, (C, reversibility), signal reversibility calculated by setting the minimumFRET ratio value equal and the initial FRET value before GIP stimulation to 100%. Values are the mean ± SEM of 8–10 records (noticed on figure). The data indicate that cAMPresponse was reversible in cells having GIPR internalization blocked whereas cAMP response was only weakly reversible in control cells internalizing GIPR.

S. Ismail et al. / Biochemical Pharmacology 120 (2016) 33–45 39

3.4. Direct identification of cAMP production on early endosomescontaining internalized GIPR

To further demonstrate production of cAMP by internalizedGIPR, we performed detection of cAMP in endosomes. For this pur-

pose, we took advantage of the fact that PtdIns(3)P is the mostabundant phosphoinositide in the membrane of early endosomesto target expression of a FRET cAMP probe at the external surfaceof early endosomes using FYVE domain of EEA1. The FYVE domainis a protein motif that allows the interaction of cytosolic proteins,

Fig. 5. Inhibition of GIPR internalization abolished irreversibility of GIP-induced PKA activation. Stable Flp-InTMGIPR-293 cells were transfected with plasmid encoding PKAbiosensor AKAR3. 24 h post-transfection, cells were washed and then incubated with medium (control curve) or with medium containing 30 lM Dyngo-4a (Dyngo-4a curve).Next, cells were superfused with 10 nM GIP for 2 min with subsequent washout. All experiments were performed at 25 �C. During these experiments, FRET was recorded asdescribed in ‘‘Materials and Methods” section. Two representative records are shown. Values of YFP/CFP FRET ratios are expressed relative to initial FRET ratio which wasnormalized to 1.00. For analysis of reversibility of PKA response after GIP washout, (panel B), signal reversibility calculated by setting the minimum FRET ratio value equal andthe initial FRET value before GIP stimulation to 100%. Values are the mean ± SEM of 8 records (noticed on figure). The data indicate that PKA activity was reversible in cellshaving GIPR internalization blocked whereas PKA activity was only weakly reversible in control cells internalizing GIPR.

0 15’ 60 ‘

GIPR-GFP

GIPR-GFP + Dyngo-4a

+ 1 μM GIP

0 15’ 60 ‘

A

0 20 40 60

20

40

60

80

100

Binding time (min)

125 I

-Phe

1 -G

IP b

indi

ng

(% m

ax.)

+ Dyngo-4a Control

B

-10 -9 -8 -7 -6

20

40

60

80

100

log [GIP], M

+ Dyngo-4a Control

C

0 20 40 600

50

100

control 30μM Dyngo-4a

cAM

P pr

oduc

tion

(1

/BR

ET

)

Stimulation time (min)

+ Forskolin

D

0 20 40 600

50

100

Stimulation time (min)

control 60μM Dyngo-4a + Forskolin

E

Fig. 6. Dyngo4-a efficiently inhibited GIPR internalization and did not affect GIP binding and foskolin-stimulated cAMP production. HEK T cells transiently expressing theGFP-tagged GIPR (termed GFP-GIPR) were stimulated with 1 lM GIP at 37� C with or without 30 lM Dyngo-4a, an inhibitor of dynamin-dependent internalization. Confocalmicroscopy images were recorded at various times of stimulation. Panel A corresponds to images of representative microscopy field at times 15 and 60 min of stimulation,illustrating kinetic of GIP-stimulated internalization of GFP-GIPR (upper images) which is inhibited by Dyngo-4a (lower images). In panels B and C, Flp-InTMGIPR-293 cellswere incubated with 125I-Phe1-GIP with or without 1 lM GIP (for non specific binding determination) in the presence or absence of 30 lM Dyngo-4a. Specific 125I-Phe1-GIPbinding was determined at various times of incubation (panel B) or Specific 125I-Phe1-GIP binding at 30 min of incubation was inhibited by increasing concentrations of GIP(panel C). Results are expressed as percent of maximum specific binding and values are mean ± SEM of 3 individual determinations. They indicate that Dyngo-4a did notsignificantly influence kinetic and affinity of GIP binding. Panels D and E show representative experiments of cAMP measurement by BRET carried out on Flp-InTMGIPR-293cells transfected with plasmid encoding Rluc-Epac1-Citrine and stimulated by Foskolin (10 lM) in the presence or the absence of Dyngo-4a (30 or 60 lM). Values of 1/BRETwere expressed as percent of maximum.

40 S. Ismail et al. / Biochemical Pharmacology 120 (2016) 33–45

20

40

60

80

100

30’ 60’

pcD

NA

5

Rab

7-D

N

pcD

NA

5

Rab

7-D

N

Stimulation time (min)

cAM

P r

espo

nse

(1/B

RE

T, %

Max

)

pcDNA5

DN-Rab7

0 20 40 60

50

100

150

200

+ 1 nM GIP

0 20 40 60

50

100

150

200

+ 10 nM GIP

pcDNA5

DN-Rab7

0 20 40 60

50

100

150

200

+ 100 nM GIP

pcDNA5

DN-Rab7

DY647-GIP

DY647-GIP

GFP-EEA1 DY647-GIP

DY647-GIP

DY647-GIP

GFP-EEA1 GFP-EEA1

GFP-EEA1 GFP-EEA1

DsRed-DN-Rab7 DsRed-DN-Rab7

A: DN-Rab7, 60 min

B: pcDNA5, 60 min C

D

DY

647-

GIP

/GFP

-EE

A1

co-

loca

lizat

ion

(%)

***

PCC = 0.49 ± 0.06

PCC = 0.68 ± 0.05

0 20 40 60 0

50

100 pcDNA5

DN-Rab7

Stimulation time (min)

+Forskolin

E

Fig. 7. Accumulation of activated internalized GIP receptors in early endosomes enhanced the sustained phase of cAMP production. Confocal microscopy images capturedafter 60 min of stimulation with 100 nM DY647-GIP of HEK cells expressing the human GIP receptor (Flp-InTMGIPR-293 cells) co-transfected with plasmid encoding GFP-EEA1(displayed in green) plus plasmid encoding DsRed tagged dominant negative of Rab7 (DsRed-DN-Rab7, displayed in blue, panel A), or with plasmid encoding GFP-EEA1 plusempty plasmid pcDNA5 (control, panel B). Pearson’s correlation coefficients –PCC- are indicated on images. Presence of DY647-GIP-labeld GIPR in early endosomes wasquantified by analyzing co-localization of DY647-GIP and GFP-EEA1. Results at times 30 and 60 min, reported in panel C, show persistent co-localization at time 60 min incells overexpressing DsRed-DN-Rab7, whereas a dramatic decrease is seen in control cells not transfected with DsRed-DN-Rab7 (p < 0.001). These data were confirmed bydetermination of co-localization using Mander’s coefficient (not shown). In panel D, Flp-InTMGIPR-293 cells were co-transfected with plasmid encoding BRET biosensor (Rluc-Epac1-citrine) plus plasmid encoding DsRed-DN-Rab7 or empty plasmid PcDNA5 as control. Cells were stimulated with GIP (10 or 100 nM) for BRET measurement of cAMP, asdescribed in ‘‘Materials and Methods” section. Blue records correspond to 1/BRET signal ratio from control cells and Red records correspond to 1/BRET signal ratio from cellsover-expressing DsRed-DN-Rab7. 1/BRET values expressed as percent of maximum in control assay are the mean ± SEM of 3 individual experiments. They account for anincrease of the sustained and late phase of cAMP production in the presence of DN-Rab7. In panel E, results indicate that over-expression of dominant negative of Rab7 doesnot modify forskolin-induced production of cAMP. Flp-InTMGIPR-293 cells co-transfected with plasmid encoding Rluc-Epac1-Citrine and DsRed-DN-Rab7 or pcDNA5 (control)were stimulated with 10 lM foskolin. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

S. Ismail et al. / Biochemical Pharmacology 120 (2016) 33–45 41

DY647-GIP GFP-Nb37 DsRed-DN-Rab7 DY647-GIP

Merge GFP-Nb37

D

0

20

40

60

80

100

pcD

NA

5

DN

-Rab

7

pcD

NA

5

DN

-Rab

7

30’ 60’

GFP

-Nb3

7/D

Y64

7-G

IP

co-

loca

lizat

ion

(%)

E

DY647-GIP

GFP-Nb37

DsRed-Rab5

Merge DY647-GIP GFP-Nb37 DsRed-Rab5

0’ 2.5’ 15’ 30’

B

0.5 1.0 1.50

20

40

60

80

100

GFP-Nb37 DsRed-Rab5 DY647-GIP

Position (μm)Rel

ativ

e flu

ores

cenc

e (%

max

) C

A

**

DY647-GIP GFP-Nb37

Merge

30’

DY647-GIP

GFP-Nb37

DsRed-Rab5

Merge

Fig. 8. Immunological detection of active Gas in endosomes containing activated GIP receptor. HEK cells expressing the human GIP receptor (Flp-InTMGIPR-293 cells)transfected with plasmid encoding the GFP-tagged nanobody specifically recognizing the active form of Gs (termed GFP-NB37, displayed in green) were stimulated withDY647-GIP. Fluorescence confocal microscopy images were captured at various times of stimulation. Images in panel A show co-localization of DY647-GIP, DsRed-Rab5 andGFP-Nb37 in endocytic vesicles. The white arrow on GFP-Nb37 images shows membrane recruitment of GFP-Nb37. A merge image of DY647-GIP and GFP-Nb37 colors isshown for the time 30 min of stimulation with DY647-GIP. It can be appreciated from an image zoom of the cell that many endosomes labeled by DY647-GIP are also labeledby the antibody against active Gas, GFP-Nb37 (displayed by white arrows). In panels B and C are shown an image zoom of an endosome and the line scan, respectively. PanelD, confocal microscopy images of Flp-InTMGIPR-293 cells co-transfected with plasmid encoding GFP-NB37, (displayed in green) and DsRed-DN-Rab7 (displayed in blue) at time60 min of stimulation with 100 nM DY647-GIP. Upper images correspond to cells co-transfected with plasmids encoding GFP-NB37 and DsRed-DN-Rab7 whereas lowerimages correspond to control cells co-transfected with plasmids encoding GFP-NB37 and pcDNA5. Panel E reports quantification of co-localization indicating that 76.0 ± 7.0%of endocytic vesicles containing DY647-GIP-bound GIPR were labeled with GFP-Nb37 at time 30 min. This ratio decreased to 48.9 ± 7.9% at 60 min. In cells over-expressingDsRed-DN-Rab7, the ratio of endocytic vesicles labeled with Nb-37-GFP was high, 82.8 ± 4.1% and 78.5 ± 4.7% at times 30 and 60 min, respectively. Co-localization ratios at60 min were significantly different in control cells and in cells over-expressing DsRed-DN-Rab7, p < 0.002. (For interpretation of the references to color in this figure legend,the reader is referred to the web version of this article.)

42 S. Ismail et al. / Biochemical Pharmacology 120 (2016) 33–45

S. Ismail et al. / Biochemical Pharmacology 120 (2016) 33–45 43

such as EEA1, with membranes containing the lipid phosphatidyli-nositol 3-phosphate [28]. The probe for cAMP detection was thus aFRET probe consisting in EYFP-Epac1 (binding domain)-ECFP-EAE1(FYVE domain). This probe (termed FYVE-EPAC) was efficiently tar-geted to the membrane of endosomes in the majority of cells(Fig. 9). In order to determine if internalized GIPR continued tostimulate production of cAMP, we compared FRET ratio in earlyendosomes containing DY647-GIP-bound GIPR which thereforecontained activated GIPR, with FRET ratio in pre-existing endo-somes lacking DY647-GIP from the same cell. As an example, onFig. 9B, are outlined images of 4 endosomes from a single cell stim-ulated with DY647-GIP showing that FRET ratios were 2.60 and2.67 at the surface of the two endosomes containing DY647-GIP-bound GIPR whereas FRET ratios were 2.91 and 3.00 at the surfaceof the two endosomes lacking DY647-GIP-bound GIPR. The lowerFRET ratio at the surface of endosomes containing DY647-GIP-bound GIPR most likely reflected endosomal production of cAMP.Indeed, the 4 endosomes were equally distant to the cell surfaceand were equally subject to diffusible cAMP produced at theplasma membrane. Furthermore, analysis of a large population ofendosomes from 13 individual cells stimulated with DY647-GIPindicated that in each cell (represented by an empty circle linked

CFP-EPAC-YFP-FYVE

GIP DY-647

CFP EVYF-PFY-CAPE-

CFP-EPAC-YFP-FYVE

A

B

2.60

2.67

3.00

2.91

Fig. 9. Direct detection of cAMP production by FRET in endosomes containing activatedDN-Rab7 and plasmid encoding FRET sensor of cAMP targeted to early endosomes thanDY647-GIP for 30 min. Panel A shows representative confocal microscopy images of cellsonly red fluorescence of DY647-GIP and green fluorescence of cAMP FRET sensor are disFRET ratio of 2.60 and 2.67 and two vesicles devoid of DY647-GIP presenting a FRET ratiothe surface of endosomes containing or not DY647-GIP-activated GIPR was performed. Pa(empty circles) or on endosomes containing DY647-GIP-bound GIPR (empty squares). Entotal of 66 and 53 vesicles were analyzed. Each couple of symbols (circle or square) linkethat in each cell, FRET ratio is lower in endosomes containing DY647-GIP-bound GIPR thato color in this figure legend, the reader is referred to the web version of this article.)

by a dotted line to an empty square), FRET ratio at the membranesurface of early endosomes containing activated GIPR was lowerthan FRET ratio in early endosomes lacking activated GIPR (Fig. 9C).Averaged FRET ratios were 2.56 ± 0.03 versus 2.83 ± 0.04, p < 0.001at time 30 min of stimulation. These data unequivocally establishthe existence of production of cAMP at the surface of endosomescontaining both GIPR and its ligand

4. Discussion

The current work provides a series of experimental datademonstrating that, once internalized and located in the mem-brane of early endosomes, GIPR continues to be activated and totrigger production of cAMP. Direct evidence for the activity ofinternalized GIPR is based on identification of the active form ofGas in early endosomes containing GIP using a genetically encodedGFP tagged nanobody, and on detection of a distinct FRET signalaccounting for cAMP production at the surface of endosomes con-taining GIP compared to endosomes without GIP. Furthermore,strong indirect experimental proofs were obtained on cells treatedwith a pharmacological inhibitor of GIPR internalization, such as

egreMGIP DY-647

C

FRE

T ra

tio (Y

FP/C

FP )

2.5

3.0

3.5

DY647-GIP + - 2.0

GIP receptors. Flp-InTMGIPR-293 cells co-transfected with plasmid encoding DsRed-ks to FYVE sequence, (termed CFP-EPAC-YFP-FYVE), were stimulated with 100 nMpresenting endocytic vesicles. Panel B shows a confocal microscopy image on whichplayed for convenience. Two vesicles containing DY647-GIP-bound GIPR presentingof 2.91 and 3.00 are pointed to illustrate how measurement of cAMP production onnel C is the representation of average FRET values on endosomes lacking DY647-GIPdosomes taken in 13 cells from 3 independent experiments and corresponding to ad by a dotted line corresponds to a single cell. From this representation it is evidentn in endosomes lacking DY647-GIP-bound GIPR. (For interpretation of the references

44 S. Ismail et al. / Biochemical Pharmacology 120 (2016) 33–45

decrease of the sustained phase of cAMP and PKA kinetics andreversibility of responses following GIP washout. Furthermore acontinuous increase of cAMP levels over time was observed in cellsexpressing genetically encoded dominant-negative of Rab7 whichcauses accumulation of early endosomes.

In fact, direct identification of active Ga and cAMP productionat the surface of endosomes was made possible by the combina-tional use of three major biological tools: the fluorescent agonistof the GIPR, DY647-GIP, the fluorescent nanobody recognizingactive Gas and FYVE-EPAC-based FRET sensor for local endosomalcAMP production. DY647-GIP, which retains biological activity ofnative GIP and which fully co-localizes with GIPR for at least thefirst 2 h of post-endocytic trafficking [18], enabled to distinguishendosomes containing activated GIPR from endosomes devoid offluorescent GIP and therefore, not containing activated GIPR. Thesecond crucial tool was the GFP-tagged nanobody, GFP-Nb37. Itspecifically recognizes the guanine-nucleotide-free form of Gasrepresenting the catalytic intermediate of Gas activation [9]. Thethird biological tool, especially developed for this study, wasFYVE-EPAC FRET sensor for cAMP which was specifically targetedto early endosomes by insertion of FYVE sequence. Indeed, FYVEsequence, derived from EEA1, provided binding of the FRET sensorto PtdIns(3)P contained in the membrane of early endosomes [28].

Our findings contradict the classical concept whereby G-proteindependent signaling of GPCRs occurs exclusively at the plasmamembrane and that internalized GPCRs are silent with respect tothis signaling pathway. On the other hand, based on these findings,the GIPR joins the group of GPCRs for which endosomal G-proteincoupled signaling have been suggested or demonstrated. So far,this group of GPCRs includes Parthyroid Hormone receptor,Tyroid-Stimulating Hormone receptor, Vasopressin 2 receptor,dopamine D1 receptor, Sphingosine-1-phosphate receptor, b2-adrenergic receptor, Pituitary Adenylate Cyclase 1 receptor andGlucagon-like peptide 1 receptor [5–9,11,29,30].

In the current study, kinetic profiles of cAMP production andPKA activity support the view that cAMP signaling resulting fromplasma membrane GIPR is subject to desensitization whereasendosomal cAMP signaling is more sustained. How internalizedGIPR triggers G-protein dependent signaling in endosomes andhow this spatio-temporal signaling can be reconciled with sequen-tial events governing GIPR activation-desensitization (inactivation)cycle at the plasma membrane are important and timely issueswhich are not elucidated yet. In the canonical cycle of Gs-coupled GPCR activation and desensitization, the agonist ligandbinds to an active form of the receptor and stabilizes it, permittingcoupling to cognate heterotrimeric GDP-bound G proteins. Theactive receptor-G-protein complex catalyzes GDP-GTP exchangecausing dissociation of the Ga subunit and bc dimer from thereceptor. Next, GTP-bound Ga binds to and activates adenylatecyclase that converts ATP into cAMP. After G-protein coupling/decoupling, the receptor is phosphorylated by GRK and by secondmessenger-dependent kinases (such as PKA), an event, which mostoften causes recruitment of the adaptor proteins b-arrestins. b-arrestins terminate G-protein-dependent signal and permit thetargeting of the receptor to clathrin-coated pits for endocytosis.At this step, the receptor is associated with a set of proteins ofinternalization and trafficking cell machineries. Studies with b2-adrenergic receptor support the view that, after dissociation ofGas from the membrane receptor and recruitment of b-arrestins,Gas is activated in endosomes containing b2-adrenergic receptorpresumably free of b-arrestins [9]. A different mechanism has beenelucidated for the PTH and vasopressin 2 receptors in whichagonist-activated receptors form a ternary complex that includesarrestins and Gbc dimer which accelerate rate of Gas activationand increase levels of activated Gas, leading to persistent produc-tion of cAMP. In this latter example, arrestins play the role of scaf-

fold protein for maintaining activated Gas associated with thereceptors [8,31]. Very recent, identification of maga-complexescontaining a GPCR simultaneously engaged with a G-protein andb-arrestin bound to phosphorylated C-terminal tail of the receptor,provided physical and biochemical rationale for the existence offunctional GPCR complexes in endosomes [32]. In the current caseof GIPR, it is unlikely that b-arrestins could play a role in activationof Gas. Indeed, we showed that the GIPR does not recruit b-arrestins upon activation [18,19] and GFP tagged b-arrestin-1 or -2 were never identified in endosomes containing GIPR [18]. Hence,the molecular partners of endosomal G-dependent signaling of theGIPR, as well as precise mechanism of plasma membrane GIPRdesensitization, remain to be identified.

Another difference between GIPR and PTHR concerns the influ-ence of endosome acidification for the termination of endosomalcAMP production. Indeed, we did not observe any significant effectof bafilomycin, an inhibitor of v-ATPase, on the sustained phase ofcAMP production (not illustrated), whereas, for the PTH receptor,based on results with this inhibitor, it has been proposed thatendosomal cAMP production is turned off by endosome acidifica-tion which dissociates PTH from its receptor. On the other hand,as for GIPR in the current study, TSHR was shown to bind TSHand be activated in acidic conditions [6].

In line with presence of active Gas in endosomes containingactivated GPCRs, a recent study shows that presence of Gas inendosomes has a function which is not restricted to stimulationof diffusible second messengers. Indeed, endosomal Gas was foundto contribute to the endocytic sorting of several GPCRs to lyso-somes. This occurs through binding to GPCR-associated bindingprotein-1 (GASP1) and dysbindin, two key proteins of the endoso-mal sorting of GPCRs in intra-luminal vesicles of multivesicularbodies [33]. Interestingly, this function of Gas in lysosomal degra-dation of GPCRs does not require its activation, supporting a scaf-folding rather a signaling function. In light of these findings, it isplausible that during post-endocytic sorting of GPCRs, Gas succes-sively plays a role in their endosomal signaling in early endosomesand then, in their trafficking to lysosomes.

Results from the current study on HEK cells represent a realmilestone and call future works on cells naturally expressing GIPR.Indeed, a major issue raised by endosomal cAMP signaling lies inits physiological relevance. In fact, the novelty of the discovery thatGPCRs continue to stimulate cAMP formation on endosomes isenhanced by accumulating data demonstrating that cAMP is com-partmentalized in cells to enable discrete pathways of localizedsignaling and related physiological responses to occur [34]. In thisregard, it can be hypothesized that endosomal cAMP productionmediated by internalized GPCRs contributes to the spatio-temporal organization of cAMP signaling cascade involving PKAs,EPACs, AKAPs, PDEs, etc.

As an example supporting this concept, it was reported thatinternalization of the TSH receptor is required to ensure normalactin rearrangement in thyroid follicles and phosphorylation ofvasodilator-stimulated phosphoprotein (VASP) [6]. As anotherexample, the duration of PTH-induced calcemic and phosphateresponses in mice which are dependent of PTH ligands, was relatedto the duration of cAMP responses in correspondence with the abil-ity of the ligands to stimulate cAMP from internalized PTH recep-tors [5]. The divergent antinatriuretic and antidiuretic actions ofvasopressin and oxytocin, both acting on V2 vasopressin receptor,were also explained by different profiles of cAMP responses to thetwo hormones, with vasopressin causing an endosomal sustainedcAMP production and oxytocin causing a short cAMP production,limited to the plasma membrane [8]. Moreover, a relationshipbetween expression of cAMP-dependent genes and endosomalcAMP production triggered by internalized b-adrenergic receptorwas recently reported [12]. In this study, inhibitors of b-

S. Ismail et al. / Biochemical Pharmacology 120 (2016) 33–45 45

adrenergic receptor internalization reduced PKA phosphorylationof CREB and expression of several cAMP-regulated genes, includingthe gene encoding the phosphoenolpyruvate carboxykinase 1. Inlight to these data, our forthcoming investigations should addressthe question of how endosomal GIP-induced cAMP signaling isrequired for full physiological functions of GIP. Such a challengingquestion requires development of new relevant biological toolssuch as endocrine pancreatic b-cells expressing the GIPR at a phys-iological level.

In conclusion, we have demonstrated that, once internalizedand located in the membrane of early endosomes, the receptorfor Glucose-dependent Insulinotropic receptor (GIPR) continuesto trigger production of cAMP which accounts for sustained cAMPproduction and PKA activation. Hence, production of endosomalcAMP by internalized presumably contributes to the spatio-temporal organization of cAMP signaling cascade downstreamGIPR.

Conflict of interest

The authors declare have no conflict of interest to declare.

Acknowledgements

The excellent technical contribution of Tobias Goldak toplasmid construction is acknowledged. Ismail Sadek received aPhD frant from Société Française d’Endocrinologie. ViacheslavO Nikolaev’s professorship is funded by Gertraud und Heinz RoseFoundation. Cellular Imaging Facility of Rangueil is granted byRégion Midi-Pyrénées. The support of COST action CM 1207 isacknowledged.

References

[1] S.S. Ferguson, Evolving concepts in G protein-coupled receptor endocytosis:the role in receptor desensitization and signaling, Pharmacol. Rev. 53 (2001)1–24.

[2] A. Sorkin, M. von Zastrow, Endocytosis and signaling: intertwining molecularnetworks, Nat. Rev. Mol. Cell Biol. 10 (2009) 609–622.

[3] S. Rajagopal, K. Rajagopal, R.J. Lefkowitz, Teaching old receptors new tricks:biasing seven-transmembrane receptors, Nat. Rev. Drug Discov. 9 (2010) 373–386.

[4] A. Marchese, M.M. Paing, B.R. Temple, J. Trejo, G protein-coupled receptorsorting to endosomes and lysosomes, Annu. Rev. Pharmacol. Toxicol. 48 (2008)601–629.

[5] S. Ferrandon, T.N. Feinstein, M. Castro, B. Wang, R. Bouley, J.T. Potts, et al.,Sustained cyclic AMP production by parathyroid hormone receptorendocytosis, Nat. Chem. Biol. 5 (2009) 734–742.

[6] D. Calebiro, V.O. Nikolaev, M.C. Gagliani, T. de Filippis, C. Dees, C. Tacchetti,et al., Persistent cAMP-signals triggered by internalized G-protein-coupledreceptors, PLoS Biol. 7 (2009) e1000172.

[7] S.J. Kotowski, F.W. Hopf, T. Seif, A. Bonci, M. von Zastrow, Endocytosispromotes rapid dopaminergic signaling, Neuron 71 (2011) 278–290.

[8] T.N. Feinstein, N. Yui, M.J. Webber, V.L. Wehbi, H.P. Stevenson, J.D. King Jr,et al., Noncanonical control of vasopressin receptor type 2 signaling byretromer and arrestin, J. Biol. Chem. 288 (2013) 27849–27860.

[9] R. Irannejad, J.C. Tomshine, J.R. Tomshine, M. Chevalier, J.P. Mahoney, J.Steyaert, et al., Conformational biosensors reveal GPCR signaling fromendosomes, Nature 495 (2013) 534–538.

[10] M. Zaccolo, T. Pozzan, Discrete microdomains with high concentration of cAMPin stimulated rat neonatal cardiac myocytes, Science 295 (2002) 1711–1715.

[11] F. Mullershausen, F. Zecri, C. Cetin, A. Billich, D. Guerini, K. Seuwen, Persistentsignaling induced by FTY720-phosphate is mediated by internalized S1P1receptors, Nat. Chem. Biol. 5 (2009) 428–434.

[12] N.G. Tsvetanova, M. von Zastrow, Spatial encoding of cyclic AMP signalingspecificity by GPCR endocytosis, Nat. Chem. Biol. 10 (2014) 1061–1065.

[13] E. Geras-Raaka, S. Neumann, M.C. Gershengorn, Persistent cAMP signaling byTSH receptors revealed by phosphodiesterase inhibition, Thyroid 23 (2013)1484–1489.

[14] S. Neumann, E. Geras-Raaka, B. Marcus-Samuels, M.C. Gershengorn, PersistentcAMP signaling by thyrotropin (TSH) receptors is not dependent oninternalization, FASEB J. 24 (2010) 3992–3999.

[15] R. Irannejad, N.G. Tsvetanova, B.T. Lobingier, M. von Zastrow, Effects ofendocytosis on receptor-mediated signaling, Curr. Opin. Cell Biol. 35 (2015)137–143.

[16] L.L. Baggio, D.J. Drucker, Biology of incretins: GLP-1 and GIP, Gastroenterology132 (2007) 2131–2157.

[17] B. Waser, R. Rehmann, C. Sanchez, D. Fourmy, J.C. Reubi, Glucose-dependentinsulinotropic polypeptide receptors in most gastroenteropancreatic andbronchial neuroendocrine tumors, J. Clin. Endocrinol. Metab. 97 (2012) 482–488.

[18] S. Ismail, I. Dubois-Vedrenne, M. Laval, I.G. Tikhonova, R. D’Angelo, C. Sanchez,et al., Internalization and desensitization of the human glucose-dependent-insulinotropic receptor is affected by N-terminal acetylation of the agonist,Mol. Cell. Endocrinol. 414 (2015) 202–215.

[19] S. Al-Sabah, M. Al-Fulaij, G. Shaaban, H.A. Ahmed, R.J. Mann, D. Donnelly, et al.,The GIP receptor displays higher basal activity than the GLP-1 receptor butdoes not recruit GRK2 or arrestin3 effectively, PLoS ONE 9 (2014) e106890.

[20] T. Yaqub, I.G. Tikhonova, J. Lattig, R. Magnan, M. Laval, C. Escrieut, et al.,Identification of determinants of glucose-dependent insulinotropicpolypeptide receptor that interact with N-terminal biologically active regionof the natural ligand, Mol. Pharmacol. 77 (2010) 547–558.

[21] S. Gremlich, A. Porret, E.H. Hani, D. Cherif, N. Vionnet, P. Froguel, et al., Cloning,functional expression, and chromosomal localization of the human pancreaticislet glucose-dependent insulinotropic polypeptide receptor, Diabetes 44(1995) 1202–1208.

[22] V.O. Nikolaev, M. Bunemann, L. Hein, A. Hannawacker, M.J. Lohse, Novel singlechain cAMP sensors for receptor-induced signal propagation, J. Biol. Chem. 279(2004) 37215–37218.

[23] M.D. Allen, J. Zhang, Subcellular dynamics of protein kinase A activityvisualized by FRET-based reporters, Biochem. Biophys. Res. Commun. 348(2006) 716–721.

[24] E. Pardon, T. Laeremans, S. Triest, S.G. Rasmussen, A. Wohlkonig, A. Ruf, et al., Ageneral protocol for the generation of Nanobodies for structural biology, Nat.Protoc. 9 (2014) 674–693.

[25] M. Foucaud, E. Marco, C. Escrieut, C. Low, B. Kalindjian, D. Fourmy, Linkingnon-peptide ligand binding mode to activity at the human cholecystokinin-2receptor, J. Biol. Chem. 283 (2008) 35860–35868.

[26] A. McCluskey, J.A. Daniel, G. Hadzic, N. Chau, E.L. Clayton, A. Mariana, et al.,Building a better dynasore: the dyngo compounds potently inhibit dynaminand endocytosis, Traffic 14 (2013) 1272–1289.

[27] E. Girard, D. Chmiest, N. Fournier, L. Johannes, J.L. Paul, B. Vedie, et al., Rab7 isfunctionally required for selective cargo sorting at the early endosome, Traffic15 (2014) 309–326.

[28] D.C. Lawe, V. Patki, R. Heller-Harrison, D. Lambright, S. Corvera, The FYVEdomain of early endosome antigen 1 is required for both phosphatidylinositol3-phosphate and Rab5 binding. Critical role of this dual interaction forendosomal localization, J. Biol. Chem. 275 (2000) 3699–3705.

[29] R.S. Kuna, S.B. Girada, S. Asalla, J. Vallentyne, S. Maddika, J.T. Patterson, et al.,Glucagon-like peptide-1 receptor-mediated endosomal cAMP generationpromotes glucose-stimulated insulin secretion in pancreatic beta-cells, Am.J. Physiol. Endocrinol. Metab. 305 (2013) E161–E170.

[30] L.A. Merriam, C.N. Baran, B.M. Girard, J.C. Hardwick, V. May, R.L. Parsons,Pituitary adenylate cyclase 1 receptor internalization and endosomal signalingmediate the pituitary adenylate cyclase activating polypeptide-inducedincrease in guinea pig cardiac neuron excitability, J. Neurosci. 33 (2013)4614–4622.

[31] V.L. Wehbi, H.P. Stevenson, T.N. Feinstein, G. Calero, G. Romero, J.P. Vilardaga,Noncanonical GPCR signaling arising from a PTH receptor-arrestin-Gbetagamma complex, Proc. Natl. Acad. Sci. U.S.A. 110 (2013) 1530–1535.

[32] A.R. Thomsen, B. Plouffe, T.J. Cahill 3rd, A.K. Shukla, J.T. Tarrasch, A.M. Dosey,et al., GPCR-G protein-beta-arrestin super-complex mediates sustained Gprotein signaling, Cell 166 (2016) 907–919.

[33] S. Rosciglione, C. Theriault, M.O. Boily, M. Paquette, C. Lavoie, Galphasregulates the post-endocytic sorting of G protein-coupled receptors, Nat.Commun. 5 (2014) 4556.

[34] K. McCormick, G.S. Baillie, Compartmentalisation of second messengersignaling pathways, Curr. Opin. Genet. Dev. 27 (2014) 20–25.