Cover Page The handle

(1)

Cover Page

The handle http://hdl.handle.net/1887/57506 holds various files of this Leiden University dissertation

Author: Fokkelman, Michiel

Title: Image-based phenotypic screening for breast cancer metastasis drug target discovery

Date: 2017-11-22

(2)

25

Chapter 3

The adhesion G protein-coupled receptor G2

(ADGRG2/GPR64) constitutively activates SRE and NFκB and is involved in cell adhesion and migration

This chapter has been published as:

Miriam C. Peeters*, Michiel Fokkelman*, Bob Boogaard, Kristoffer L. Egerod, Bob van de Water, Ad P. IJzerman, Thue W. Schwartz

* both authors contributed equally

The adhesion G protein-coupled receptor G2 (ADGRG2/GPR64) constitutively activates SRE and NFκB and is involved in cell adhesion and migration Cellular Signalling, Volume 27, December 2015

(3)

Chapter 3

26

Abstract

Adhesion G protein-coupled receptors (ADGRs) are believed to be activated by auto-proteolytic cleavage of their very large extracellular N-terminal domains normally acting as a negative regulator of the intrinsically constitutively active seven transmembrane domain. ADGRG2 (or GPR64) which originally was described to be expressed in the epididymis and studied for its potential role in male fertility, is highly up-regulated in a number of carcinomas, including breast cancer. Here, we demonstrate that ADGRG2 is a functional receptor, which in transfected HEK293 cells signals with constitutive activity through the adhesion- and migration-related transcription factors serum response element (SRE) and nuclear factor kappa-light-chain-enhancer of activated B cells (NFκB) presumably via coupling to Gα12/13 and Gαq. However, activation of these two pathways appears to occur through distinct molecular activation mechanisms as autoproteolytic cleavage is essential for SRE activation but not required for NFκB signaling. The overall activation mechanism for ADGRG2 is clearly distinct from the established ADGR activation mechanism as it requires the large extracellular N-terminal domain for proper intracellular signal transduction. Knockdown of ADGRG2 by siRNA in the highly motile breast cancer cell lines Hs578T and MDA-MB-231 resulted in a strong reduction in cell adhesion and subsequent cell migration which was associated with a selective reduction in RelB, an NFκB family member. It is concluded that the adhesion GPCR ADGRG2 is critically involved in the adhesion and migration of certain breast cancer cells through mechanisms including a non-canonical NFkB pathway and that ADGRG2 could be a target for treatment of certain types of cancer.

(4)

GPR64 constitutively activates SRE and NFκB and is involved in cell adhesion and migration

27

1. Introduction

The family of G protein-coupled receptors (GPCRs) consist of over 800 distinct members divided into six subfamilies (A-F) of which the class A or Rhodopsin-like is the largest and most extensively studied [1]. The class B or secretin-like subfamily can be further divided in two different families;

the B1 hormone receptors and the B2 adhesion GPCRs. ADGRG2/GPR64 is part of the latter group, the largest orphan-containing receptor family of GPCRs, i.e. for the majority of adhesion GPCRs the endogenous effector molecule has not yet been identified [2]. Also ADGRG2/GPR64 is still a full orphan with no synthetic or endogenous ligands known.

ADGRG2 was first identified by differential screening of a human epididymal cDNA library and was therefore given the name human epididymis-specific protein (HE6) [3]. Since the identified 5kb mRNA encoded a G protein-coupled receptor, the HUGO Gene Nomenclature Committee later changed the name to GPR64. In the beginning of 2015, the IUPHAR accepted a new, uniform nomenclature for the family of adhesion GPCRs and according to this new nomenclature, the receptor was renamed Adhesion G protein-coupled receptor subfamily G member 2 (ADGRG2) [4].

ADGRG2 has a classical adhesion GPCR structure, with seven transmembrane helices and a large N-terminus containing a GPCR Proteolytic Site (GPS) that can be autoproteolytically cleaved while being part of a larger domain, the GPCR Autoproteolysis Inducing (GAIN) domain that is sufficient and essential for proper cleavage [5].

In healthy human tissue, ADGRG2 has only been reported to be highly expressed in the epididymis, where it has been suggested to play a role in regulating sperm fluid homeostasis [6, 7]. Knockout ADGRG2 male mice proved to be infertile, which appeared related to spermatozoa stability and fluid back-up in the testis [7]. It is unclear whether ADGRG2 plays an equally important role in human fertility.

ADGRG2 has also been found to be upregulated in various carcinomas, including prostate, kidney, breast and non-small cell lung cancer. Furthermore, knockdown of ADGRG2 in Ewing sarcoma suppressed tumor growth and metastasis formation in mice [8].

Several adhesion GPCRs have been shown to be involved in cell adhesion and migration, hereby influencing tumor progression. For example, loss of ADGRG1 (GPR56) decreases granule cell adhesion [9] and cellular adhesion of acute myeloid leukemia cells [10]. Furthermore, Iguchi et al. published that ADGRG1 is a regulator of neuronal progenitor cell migration by activating Gα12/13 and Rho-dependent transcription factors [11]. ADGRF5 (GPR116) knockdown inhibited breast cancer cell migration and invasion both in vitro and in vivo via activation of the RhoA and Rac1 pathways [12]. Overexpression of ADGRE5 (CD97) in HT1080 cells increased single cell random migration in vitro and promoted tumor growth in scid mice [13].

(5)

Chapter 3

28

During the rather short history of adhesion GPCR research, several activation mechanisms have been proposed [14]. However, the last few years more adhesion GPCRs are described that show a similar way of activation, leading to the hypothesis that there is a general activation mechanism for most, if not all members of the family [15]. Adhesion GPCRs generally undergo autoproteolysis separating the large N-terminus and the seven transmembrane (7TM) domain.

Yet, the two protein domains stay in close proximity of each other [5, 16]. In the general activation mechanism, the N-terminal domain functions as a negative regulator, suppressing the signaling ability of the 7TM domain. Only when the N-terminus dissociates from the 7TM domain, the receptor can transmit its full biological signal. Removal of the N-terminus either releases the intrinsically constitutive active 7TM domain [17], or it allows binding of an otherwise shielded agonist [18, 19]. Support for this general activation mechanism has been described for ADGRB2 (BAI2) [20], ADGRG1 (GPR56) [21], ADGRE5 (CD97) [22], ADGRF1 (GPR110) [19], ADGRD1 (GPR133) and ADGRG6 (GPR126) [18]. Although for a number of adhesion GPCRs it has now been acknowledged that activation often includes G protein coupling [4, 15], no signaling ability of ADGRG2, whether G protein-dependent or independent, has been demonstrated so far.

Here we provide evidence that in contrast to the hypothesized general activation mechanism of adhesion GPCRs, the N-terminus of ADGRG2 acts as positive regulator important to maintain a conformation that allows constitutive activity of the receptor. Removal of the N-terminal domain results in a complete loss of activation. Furthermore, we show that ADGRG2 is involved in both cell adhesion and migration, possibly via Gα12/13 and Gαq induced activation of transcription factors serum response element (SRE) and nuclear factor kappa-light-chain-enhancer of activated B cells (NFκB), respectively. However, ADGRG2 activates these pathways differentially, displaying constitutive biased signaling.

2. Materials and Methods

2.1 Cell culture and transient transfection

HEK293 cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FCS (Sigma Life Science, Saint-Louis, CA, USA) at 37 °C in a humidified 7% CO2 incubator.

Cells were transiently transfected with 20-40 ng/well (96-well) or 1 µg/well (6-well) hADGRG2 constructs using Lipofectamine 2000 (Life technologies, Carlsbad, CA, USA) according to the manufacturer's recommendations. For SRE and NFκB luciferase assays, cells were co-transfected with 50 ng/well of the cis-reporter plasmids pSRE-luc or pNFκB-luc (PathDetect, Stratagene, Ja Jolla, CA, USA).

Hs578T and MDA-MB-231 cells were grown in RPMI-1640 medium (Life technologies, Carlsbad, CA, USA) supplemented with 10% FCS at 37 °C in a humidified 5% CO2 incubator. Stable GFP- expressing Hs578T and MDA-MB-231 cells were generated by lentiviral transduction of pRRL- CMV-GFP and selection of GFP positive clones by FACS.

(6)

29 Transiently siRNA knockdown was achieved by reverse transfection of 50nM single or smart pool siRNA (Dharmacon, Lafayette, CO, USA) in 7,500 cells/well in a 96-well plate format using the transfection reagent INTERFERin (Polyplus, Illkirch, France) according to the manufacturer’s guidelines. Medium was refreshed after 20 h and transfected cells were used for various assays between 50 to 72 h after transfection.

2.2 Phagokinetic track assay

Black 96-well µClear plates (Greiner Bio-One, Frickenhausen, Germany) were coated with 10 µg/ml fibronectin (Sigma Aldrich, Saint-Louis, CA, USA) for 1 h at 37°C. Plates were washed twice with PBS, using a HydroFlex platewasher (Tecan, Männedorf, Switzerland). Subsequently, the plates were coated with white carboxylate modified latex beads (400 nm, 3.25·10⁹ particles per well; Life Technologies, Carlsbad, CA, USA) for 1 h at 37°C, after which the plate was thoroughly washed with PBS. 65 h after siRNA transfection, single cell suspensions were seeded at low density (~100 cells/well) in the beads-coated plate. Cells were allowed to migrate for 7 h, after which the cells were fixed for 10 min with 4% formaldehyde. Migratory tracks were visualized by acquiring whole well montages (6x6 images) on a BD Pathway 855 BioImager (BD Biosciences, Franklin Lakes, NJ, USA) using transmitted light and a 10x objective (0.40 NA). Montages were analyzed using WIS PhagoTracker [23] and quantitative data of at least three independent experiments was analyzed using GraphPad Prism 6.0 (GraphPad Software, San Diego, CA).

2.3 Live cell migration and adhesion

Hs578T-GFP and MDA-MB-231-GFP cells were transfected with siRNAs as described above and after 65h, knockdown cell suspensions were seeded in fibronectin-coated black 96-well glass plates (SensoPlate, Greiner Bio-One, Frickenhausen, Germany). Live microscopy was performed on a Nikon Eclipse Ti microscope, equipped with a 37°C incubation chamber with CO2 flow, an automated xy-stage, Perfect Focus System and 20x objective (0.75 NA, 1.00 WD). Images were captured using a DS-Qi1MC CCD camera with 2x2 binning (pixel size: 0.64µm). To visualize cell adhesion after siRNA knockdown, Differential Interference Contrast (DIC) images were acquired directly after seeding cells. Images were acquired every minute for 2 h using NIS software (Nikon, Amsterdam, The Netherlands). For live cell migration assays, knockdown and control cells were allowed to adhere for 10 h before imaging started. siRNA targeting the GTPase dynamin 2 (DNM2) was used as positive control for reduced cell migration [24]. Two positions per well were selected and GFP images were acquired every 12 min for a total imaging period of 12 h using NIS software (Nikon). Image analysis was performed using CellProfiler (Broad Institute, [25]). Briefly, images were segmented using an in-house developed watershed masked clustering algorithm [26], after which cells were tracked based on overlap between frames. Tracking data was organized and analyzed using in-house developed R-scripts [27] to obtain single cell migration

(7)

Chapter 3

30

data. Single cell migration speeds were plotted using GraphPad Prism 6.0 and changes in migration speed were evaluated by comparing cell populations.

2.4 xCELLigence experiments

Hs578T were transfected with smart pool ADGRG2 siRNA as described above. 50 h post transfection, cells were harvested and pooled and spread in a density of 1·10⁴ cells/well on uncoated 16-well E-plates containing imbedded gold electrodes (Roche Applied Science, Mannheim, Germany). Cell attachment and proliferation changes the local ionic environment at the electrode-solution interface, thereby generating impedance, expressed as Cell Index (CI).

After cell seeding, the E-plate was placed directly into the xCELLigence RTCA system (Roche Applied Science, Mannheim, Germany) and impedance was measured for 24 h.

Data obtained from four independent experiments were captured with RTCA Software 1.2 (Roche Applied Science, Mannheim, Germany) and subsequently exported and analyzed using GraphPad Prism 6.0 (GraphPad Software Inc., San Diego, CA, USA). For data analysis, ligand responses were normalized to Δ Cell Index (ΔCI) after subtracting baseline (t=0).

2.5 Human ADGRG2 constructs and site-directed mutagenesis

Human ADGRG2 (Genebank Accession #NM_001079858) full length and deletion constructs were amplified by PCR and provided with a C-terminal rho-1D4-tag (TETSQVAPA)[28]. These constructs were subsequently cloned into the EcoRI-XbaI restriction sites of the modified vector pcDNA3.1(+), kindly provided by Kate Hansen (7TM-Pharma, Lyngby, Denmark), which contains an upstream sequence encoding a hemagglutinin signal peptide fused to a M1-FLAG tag [29].

Constructs were confirmed by double stranded sequencing (LGTC, Leiden, The Netherlands).

Forward Primers:

ADGRG2-FL (1-1017 aa)

CACCAGGAATTCTATGGTTTTCTCTGTCAGGCAG EcoRI

ADGRG2-7TM (607-1017 aa)

CATCAGGAATTCTACAAGCTTCGGCGTTCTGCTG EcoRI

ADGRG2-GPS (565-1017)

CATCAGGAATTCTACAGTGAGATGTGTATTTTGGG EcoRI

ADGRG2- GAIN (339-1017 aa)

CATCAGGAATTCTCCTGTGAAAGCCTCATTTTC EcoRI

ADGRG2-NBD (116-1017)

CATCAGGAATTCCAATGACTCAGCATTTTTTAGA EcoRI

Reverse Primer:

ADGRG2-1D4 CATCAGTCTAGATCAGGCGGGGGCCACCTGGGAGGTCTCGGTCATTTGCTCAATAAAGTG XbaI stop rho-1D4

(8)

31 Mutations T607A and S608A in constructs ADGRG2-FL and ADGRG2-GAIN were created based on the QuikChange method (Agilent Technologies, La Jolla, CA, USA) [30] using pfu polymerase (Promega, Madison, WI, USA) in an 18-cycle mutagenic PCR. Template DNA was subsequently digested by DpnI (New England Biolabs, Ipswich, MA, USA) treatment and PCR products were transformed in chemically competent DH5α cells (Life Technologies, Carlsbad, CA, USA), purified and confirmed by double stranded sequencing (LGTC, Leiden, The Netherlands).

Forward primers:

ADGRG2-T607A GTACCTGTAGCCATCTAGCAAGCTTCGGCGTTCTG

ADGRG2-S608A ACCTGTAGCCATCTAACAGCCTTCGGCGTTCTGCTG

Reverse primers:

ADGRG2-T607A CAGAACGCCGAAGCTTGCTAGATGGCTACAGGTAC

ADGRG2-S608A CAGCAGAACGCCGAAGGCTGTTAGATGGCTACAGGT

2.6 ELISA

HEK293 cells were transiently transfected as described above. After 48 h, cells were washed twice with PBS, fixed for 10 min with 3.7% formaldehyde, and incubated in Dulbecco’s Phosphate Buffered Saline (Gibco; Life Technologies, Carlsbad, CA, USA) with 2% BSA for 30 min at room temperature. To visualize the C-terminal rho-1D4 tag, cells were permeabilized by adding 0.2%

saponin (Sigma-Aldrich, Saint-Louis, CA, USA) to the blocking solution. Subsequently, the cells were incubated with 1:2250 anti-M1-FLAG antibody (Sigma-Aldrich, Saint-Louis, CA, USA) or 1:3000 rho-1D4 antibody (The University of British Colombia, Canada). The cells were washed three times with PBS and incubated with 1:1000 anti-mouse horseradish peroxidase-conjugated antibody (Sigma-Aldrich, Saint-Louis, CA, USA). Immunoreactivity was visualized by the addition of 3,3’,5,5’-Tetramethylbenzidine (TMB) (Sigma-Aldrich, Saint-Louis, CA, USA) and after 5 min, the reaction was stopped by the addition of 0.2M H2SO4. The absorbance was read at 450 nm using a VICTOR² plate reader (PerkinElmer, Groningen, The Netherlands). Data obtained from at least three independent experiments was analyzed using GraphPad Prism 6.0 (GraphPad Software, San Diego, CA).

2.7 SRE- and NFkB-luciferase assay

HEK293 cells were transiently transfected as described above with the exception that after 6 h, transfection medium was replaced with low serum growth medium (0.5% FCS).The assay was terminated 30 h post transfection by addition of luciferase assay reagent (SteadyLite;

PerkinElmer Life Sciences, Waltham, MA, USA). Luminescence was measured using the EnVision Multilabel plate reader (PerkinElmer, Groningen, The Netherlands).

When measuring the effect of inhibitors, the compounds Y27632 (25µM) and U0126 (10µM) (Sigma-Aldrich, Saint-Louis, CA, USA) were added 5 h post transfection and incubated overnight.

The compounds LY294002 (50µM) (Sigma-Aldrich, Saint-Louis, CA, USA) and UBO-QIC (1µM)

(9)

Chapter 3

32

(Institute for Pharmaceutical Biology, University of Bonn) were added 25 h after transfection and incubated for 5 h before measuring luminescence.

Data obtained by at least three independent experiments was analyzed using GraphPad Prism 6.0 (GraphPad Software, San Diego, CA).

2.8 Western Blot

HEK293 and Hs578T cells were transiently transfected as described above. Pathway inhibitors BMS-345541 (2uM, Sigma-Aldrich, Saint-Louis, CA, USA) and U0126 (10uM, Promega, Madison, WI, USA) were used as positive controls for NFkB and ERK signalling, respectively. Cells were harvested in RIPA lysis buffer (Millipore, Temecula, CA, USA) supplemented with protease inhibitors (Complete mini tablets; Roche, Mannheim, Germany. Protein concentration was determined using the Pierce BCA protein assay kit (Thermo Scientific, Rockford, IL, USA) and 25 μg protein was loaded on a 3-8% Tris-Acetate gel (HEK293) (NuPAGE/NOVEX; Life Technologies, Carlsbad, CA, USA) or 10% SDS-polyacrylamide gel (Hs578T). Protein was transferred to a PVDF membrane (Life Technologies, Carlsbad, CA, USA) by semi-dry blotting. HEK293 blots were incubated with anti-M1-FLAG antibody (1: 2500; Sigma-Aldrich, Saint-Louis, CA, USA) and anti- mouse horseradish peroxidase-conjugated antibody (1:5000; Sigma-Aldrich, Saint-Louis, CA, USA) subsequently. Chemiluminescence detection was performed by enhanced chemiluminescence substrate (Thermo Scientific, Rockford, IL, USA) and visualized on a FluorChemE digital imaging camera (Protein Simple, Santa Clara, CA, USA).

For ADGRG2 downstream signaling, Hs578T blots were probed for total ERK1/2, phospho-ERK1/2 (Thr202/Tyr204) and phospho-NFκB p65 (Ser536) (1:1000; Cell Signaling, Bioké, Leiden, The Netherlands), p65 and RelB (1:1000; Santa Cruz, Dallas, Texas), and α-tubulin (1:5000; Sigma Aldrich, Saint-Louis, CA, USA). Blots were incubated with HRP-conjugated anti-mouse and anti- rabbit antibodies (1:2000; Jackson, West Grove, PA, USA ) or with Alexa-647 conjugated anti- mouse antibody (1:5000; Jackson, West Grove, PA, USA) for tubulin. Chemiluminescence detection was performed by enhanced chemiluminescence substrate (Thermo Scientific, Rockford, IL, USA) and both ECL signal and fluorescence was visualized on a ImageQuant LAS4000 (GE Healthcare, Piscataway, NJ, USA). The results of two independently performed exeriments were normalized to mock transfected Hs578T lysates after correction for the loading control α- tubulin.

2.9 Statistical Analysis

Normality of migration measurements was tested using Kolmogorov–Smirnov’s test, d’Agostino and Pearson’s test and Shapiro–Wilk’s test using GraphPad Prism 6.0 (GraphPad Software, San Diego, CA). A data set was considered normal if found as normal by all three tests. Data sets following a normal distribution were compared with Student’s t-test (two-tailed, equal variances) using GraphPad Prism 6.0. Data sets that did not follow a normal distribution were compared

(10)

33 using Mann–Whitney’s test or a non-parametric ANOVA (Kruskal–Wallis with Dunn’s multiple comparisons post-test) using GraphPad Prism 6.0.

Statistical analysis on the signaling data was performed using Student’s t-test (two-tailed, equal variances) as available in GraphPad Prism 6.0.

3. Results

3.1 The adhesion GPCR ADGRG2 (GPR64) is involved in cell migration

To examine the influence of ADGRG2 (GPR64) on cell migration, the highly motile breast cancer cell line Hs578T was transiently transfected with a smart pool of ADGRG2 siRNA. Migratory tracks of the ADGRG2 knockdown cells were compared to mock-transfected cells in a phagokinetic track assay, allowing quantitative phenotypic assessment of cell migration behavior [31]. Here, cells were allowed to migrate on plates coated with latex beads that are removed by phagocytosis, leaving a visual track of the migratory paths of individual cells. Upon knockdown of ADGRG2, the migratory tracks displayed a smaller, more round phenotype compared to the mock situation in Hs578T cells (Fig. 1A). An analysis of all 33 human adhesion GPCR members revealed that ADGRG2 knockdown affected migration most severely when taking both the axial ration and net area of the migratory tracks into account (Fig. 1B). Quantification of the total track area and axial ratio showed a significant decrease in size and direction of the migratory paths of ADGRG2 knockdown cells (Fig. 1C). Hs578T mock cells created long tracks of 13631 ± 500 µm² with an axial ratio of 1.97 ± 0.06, whereas ADGRG2 knockdown tracks were smaller (9914 ± 650 µm²) and round (axial ratio 1.54 ± 0.06). Furthermore, migratory tracks of ADGRG2 knockdown cells scored lower for the roughness parameter, suggesting that cells have reduced membrane ruffling and form less filopodia.

The effect of ADGRG2 knockdown on Hs578T cell migration was subsequently evaluated in a live cell migration assay, where cells were continuously monitored for migration behavior for a 12 h period after full cell adherence. Consistent with the phagokinetic track assay, ADGRG2 knockdown cells appear to be stationary (Fig. 2A). Individual cells were tracked over time and siADGRG2 cells were attenuated in their migratory behavior compared to mock transfected cells almost to the same degree as the positive control dynamin 2 (DNM2) (Fig. 2B and movie S1). The GTPase DNM2 was previously described to be important for migration in cells with a focal adhesion phenotype [24]. This effect was seen for all four different single siRNAs present in the smartpool (Fig. 2C). Importantly, comparable results were obtained for another breast cancer cell line: both smartpool and individual siRNAs targeting ADGRG2 attenuated the cell migration behavior of MDA-MB-231 cells (Fig. S2).

(11)

Chapter 3

34

Figure 1. Reduced migration of Hs578T-ADGRG2 siRNA knockdown cells in Phagokinetic track assay (PKT).

A. Migratory tracks of mock and ADGRG2 siRNA transfected Hs578T cells after 7 h of migration on carboxylate modified latex beads. A representative single field of view is shown. B. Migratory tracks of cells with siRNA knockdown of all adhesion GPCRs (in grey) were quantified and normalized to tracks of Hs578T mock cells (in blue).

Average Z-score of track area and axial ratio of 4 replicates is shown. Knockdown of ADGRG2 (in red) has a strong effect on both parameters of cell migration. C. Whole well montage averages of total track area, axial ratio and roughness of >500 cells from three independent experiments. Mean ± 95% confidence interval is shown (**** = p≤0.0001 in student t-test).

3.2 ADGRG2 knockdown affects cell adhesion, but not proliferation

To assess the involvement of the adhesion GPCR ADGRG2 in adhesion, mock transfected and siRNA transfected cells were spread onto fibronectin coated glass microscopy plates and adhesion was monitored. As shown in Figure 3A, adhesion was much delayed when ADGRG2 is knocked down in Hs578T cells. Where mock transfected cells readily start to adhere within 5 minutes, ADGRG2 knockdown cells only slowly start to adhere after 10 minutes (movie S3). This

(12)

35 delayed effect on cell adhesion by siADGRG2 was also observed in MDA-MB-231 cells (Fig. S4).

Interestingly, knockdown of positive control DNM2 reduced cell migration (Fig. 2C), but did not affect cellular adhesion, suggesting that the delay in adhesion is specific for loss of ADGRG2 (data not shown). Additionally, the adherence effect was confirmed by another experimental set-up with xCELLigence Technology (Fig. 3B). This label-free technology measures cell adhesion, proliferation and responses by monitoring impedance that is generated by morphological changes of the cells on gold electrodes imbedded in the assay plates. In contrast to the microscopy plates, the E-plates do not need to be coated with fibronectin for proper adherence of wild type Hs578T cells, allowing to eliminate any influence of this coating on the observed effect on adhesion. On the E-plates, mock transfected cells are fully adhered to the plate surface in 1.47 ± 0.16 h, while siRNA transfected cells need 3.04 ± 0.20 h to reach the same plateau. After full adherence, cells will start to proliferate, resulting again in an increase of impedance measured by the xCELLigence. Interestingly, both mock transfected and ADGRG2 knockdown cell lines show an identical proliferation curve, reaching a plateau after 20 hours.

Figure 2. siRNA knockdown of ADGRG2 in Hs578T reduced single cell migration speed and distance.

A. 2D migration of Hs578T mock and ADGRG2 knockdown cells was followed by live cell microscopy. The segmentation outline of individual cells is color-coded based on time. Time points after 0, 2, and 4 hours (t=0, t=2 and t=4) are shown. B. Single cell migration trajectories were plotted for individual movies. Only xy-coordinates of cells that were tracked for the full length of the movie were used. C. Quantification of single cell migration speed after loss of ADGRG2 by single siRNA sequences and smartpool knockdown. Hs578T mock cells migrate 28.07 ± 1.71 µm/h (mean ± 95% confidence interval), whereas migration speed of ADGRG2 smartpool knockdown cells is reduced to 14.94 ± 2.28 µm/h. Smartpool siRNA is annotated as si-ADGRG2 sp and single sequences are indicated by #1-4.

Knockdown of dynamin 2 (DNM2) was used as positive control. Graph shows single cell populations with mean ± 95% confidence interval and statistical significance was determined using Kruskal-Wallis test with Dunn’s post correction (*** = p=0.0001, **** = p<0.0001). Data shown represents one replicate of three independent experiments.

(13)

Chapter 3

36

Figure 3. ADGRG2 knockdown affects Hs578T adhesion, but not proliferation

A. Cell adhesion of Hs578T mock and ADGRG2 knockdown cells was monitored by DIC live microscopy. Three independent experiments were performed, providing 30 movies per condition. Representative images of one movie are shown. B. Adhesion and proliferation of mock and ADGRG2 siRNA transfected Hs578T cells measured using the impedance-based xCELLigence technology. Mock transfected cells adhere on uncoated E-plates in 1.47 ± 0.16 h, while ADGRG2 siRNA transfected cells need 3.04 ± 0.20 h to reach full adherence. After adherence, both mock and siRNA transfected cells show similar proliferation, reaching full confluence after 20 hrs. The graph represents the combined results of four independent experiments performed in duplicate. Impedance is expressed as baseline corrected Cell Index (ΔCI).

3.3 ADGRG2 constitutively activates transcription factors SRE and NFκB

To date, ADGRG2 is still a full orphan, i.e. no ligands, synthetic or endogenous have been identified. To investigate its signaling profile, we therefore created several deletion constructs, removing the N-terminus until each of the identified domains NBD (N-terminal binding domain), GAIN, GPS and the 7 transmembrane domain (7TM) downstream from the autoproteolytic cleavage site. In Figure 4A a schematic representation of the deletion constructs is depicted.

Additionally, all constructs were N-terminally labelled with a M1-FLAG-tag and C-terminally with a Rho-1D4-tag. To promote cell surface expression, we also included an artificial signal peptide directly upstream of the FLAG-tag [29]. These tags allow for the detection of whole cell and cell surface expression by ELISA (Fig. 4B). Since full length ADGRG2 (ADGRG2-FL) also possesses a native signal peptide both the artificial signal peptide and FLAG-tag will be cleaved off, making it impossible to measure cell surface expression with the anti-FLAG-tag antibody. In a

(14)

37 permeabilized whole cell ELISA it was shown that the full length receptor is expressed in transiently transfected HEK293 cells at levels comparable to the deletion construct ADGRG2- NBD. Constructs ADGRG2-NBD, ADGRG2-GPS and ADGRG-7TM were expressed on the cell surface in comparable levels, while deletion construct ADGRG2-GAIN was more highly expressed, reaching levels even higher than the positive control Ghrelin receptor (GSHR), a class A receptor that is readily expressed after transient transfection.

Figure 4. Intracellular signaling of full length and deletion constructs of hADGRG2 in HEK293 cells

A Schematic representation of deletion constructs of human ADGRG2. FL = Full Length, NBD = N-terminal Binding Domain, GAIN = GPCR Autoproteolysis INducing domain, GPS = GPCR Proteolysis Site, 7TM = 7 Transmembrane domain. B. Expression of full length (ADGRG2-FL) and deletion constructs measured by whole cell ELISA using antibodies for either FLAG-tag located at the N-terminus or Rho-1D4 tag located at the C-terminus after cell permeabilization. Expression is shown relative to the positive control Ghrelin receptor (GSHR). C. Activation of transcription factors serum response element (SRE) and nuclear factor kappa B (NFκB) by ADGRG2 constructs.

Activation is shown relative to response of mock transfected HEK293 cells. Statistical analysis represents significant activity compared to mock situation (*** = p≤0.001 in students t-test). D. Activation of full length ADGRG2 in response to the inhibitors Y27632 (RhoA inhibitor), U0126 (ERK inhibitor), LY294002 (PI3K inhibitor) and UBO-QIC (Gαq inhibitor) is shown relative to the situation without inhibitor. (** = p≤0.01 *** = p≤0.001 in students t-test).

All bar graphs are the combined result of at least three independent experiments performed in triplicate (Mean ± SEM).

HEK293 cells transiently expressing the ADGRG2 constructs were used to investigate constitutive activity of two transcription factors, SRE and NFκB that are known to play a role in cell adhesion and migration and to be activated downstream of G protein signaling (Fig. 4C). Full length ADGRG2 has constitutive activity in both SRE and NFκB pathways of 2.5 ± 0.4 and 1.9 ± 0.2 fold over mock transfected cells respectively.

(15)

Chapter 3

38

The deletion construct ADGRG2-NBD in which the first N-terminal 116 amino acids were deleted was also constitutively active in both assays, although to a much greater extent in activating SRE with 27.8 ± 3.0 fold over mock versus 1.8 ± 0.3 fold over mock NFκB activation. ADGRG2-GAIN was able to highly constitutively activate SRE (11.5 ± 1.2), but did not show any constitutive activity of NFκB. The constructs in which we removed most of (ADGRG2-GPS) or the complete N- terminus (ADGRG2-7TM) lost all constitutive activation of both transcription factors.

To gain information of the upstream signaling pathways and G proteins involved in both SRE and NFκB activation, we applied several small molecule pathway inhibitors (Fig. 4D). Y27632 inhibits Rho kinase located downstream of Gα12/13 activation. This inhibitor, at a concentration of 25 μM, was able to reduce constitutive SRE activation of full length ADGRG2 with 77 ± 11%, indicating that a large part of the SRE activity originates from Gα12/13 coupling. The PI3K inhibitor LY294002 (50 μM), downstream of βγ-subunit signaling, also reduced SRE activity with 64 ± 15%. Both signaling pathways can result in ERK activation and when applying the MEK1/2 inhibitor U0126 (10 μM), SRE activation was almost completely abolished (inhibition of 94 ± 2%).

The constitutive activation of transcription factor NFκB by ADGRG2 appears to originate from coupling to Gαq and βγ-subunit signaling with NFκB inhibition of 60 ± 7% in the presence of the Gαq inhibitor UBO-QIC and 61 ± 9% when the PI3K inhibitor LY294002 was used.

3.4 SRE activation is dependent on ADGRG2 autoproteolysis in contrast to NFκB activation Additionally, we examined the importance of the autoproteolytic cleavage of ADGRG2. For this purpose, we created the cleavage deficient mutation T607A, and control mutation S608A that should not affect cleavage of the receptor in both ADGRG2-FL and ADGRG2-GAIN. The Western blot in Figure 5A shows the mutants in ADGRG2-GAIN of which T607 indeed disrupted cleavage with bands representing the full construct at ca. 70 kDa and 90 kDa. ADGRG2-FL and both mutants were tested for constitutive NFκB activation (Fig. 5B). All three receptors were able to activate NFκB to comparable levels. In contrast, the cleavage deficient mutation T607A significantly reduced SRE activation of both ADGRG2-FL and ADGRG2-GAIN. The control mutation S608A did not have a decreased constitutive activity and the activity was even increased when the mutation was introduced in ADGRG2-GAIN (Fig. 5C).

3.5 The NFκB family member RelB is affected by ADGRG2 knockdown

Next, we wished to address whether SRE and NFκB activation by ADGRG2 could be responsible for the observed effect on cell migration and adhesion. We performed immunostaining on Hs578T cell lysates for pERK, that was mainly responsible for the SRE signal as shown by the ERK inhibitor U0126 (Fig. 4D), and for two NFκB family members RelA/p65 and RelB. No effect was seen on pERK upon ADGRG2 knockdown, also not under starvation conditions (i.e. deprivation of serum) (Fig. 6A). Interestingly, siRNA knockdown of ADGRG2 in Hs578T cells did reveal a downregulation of the NFκB family member RelB, but not RelA (Fig. 6B).

(16)

39 Figure 5. SRE activation is dependent on ADGRG2 autoproteolysis in contrast to NFκB activation

A. ADGRG2-GAIN mutant receptor T607A cannot undergo autoproteolytic shown by immunoblotting using the anti-FLAG antibody. Cleaved ADGRG2-GAIN appears on the membrane at ca. 50 kDa and 72 kDa in lanes ADGRG2-GAIN and ADGRG2-GAIN- S608A, whereas uncleaved protein is seen at ca. 70 kDa and 90 kDa in lane ADGRG2- T607A. B. Activation of transcription factor nuclear factor kappa B (NFκB) by full length ADGRG2 and the effect of cleavage deficient mutant T607A and control mutant S608A. C. Activation of transcription factor serum response element (SRE) by full length ADGRG2 (left) and ADGRG2-GAIN (right) and by cleavage deficient mutant T607A and control mutant S608A.

Activation is shown relative to response of mock transfected HEK293 cells. All bar graphs are the combined result of at least three independent experiments performed in triplicate (Mean ± SEM). Statistical analysis represents significant activity compared to mock situation (*** = p≤0.001, **** = p≤0.0001 in students t-test). The bar graphs are the combined result of at least three independent experiments performed in triplicate.

Figure 6. ADGRG2 activates non-canonical NFκB signaling in Hs578T cells

A. Protein expression and phosphorylation of ERK1/2 in Hs578T cells is not affected by knockdown of ADGRG2.

Serum starvation is not sufficient to reduce ERK signaling, whereas knockdown of ERK2 or exposure to U0126 (10 µM for 6h) inhibits ERK phosphorylation. B. RelB expression is reduced after ADGRG2 knockdown, whereas RelA and phosphorylation of RelA are unaffected. The IKK-inhibitor BMS-345541 (2 µM for 6h) was used as a positive control for inhibition of NFκB signaling. The right panel represents quantification of the relative RelB expression of two independent experiments.

(17)

Chapter 3

40

4. Discussion

The family of adhesion G protein-coupled receptors (ADGRs/class B2 GPCRs), is one of the most ancient GPCR families, found in the most primitive animals and even among several basal fungi, and are likely ancestral to the Secretin/class B1 subfamily of GPCRs [32, 33]. ADGRs are generally very large transmembrane proteins encoded by a large number of exons. The receptors have unusually long extracellular N-terminal domains that consist of different adhesion-like domains, a high concentration of serine and threonine residues and many N- and O-glycosylation sites [34].

Partly because of this complex structural biology, information on ADGR signaling and physiological function is very limited to date. However, during the last few years, it is becoming apparent that this GPCR family plays a role in several essential cell biological processes, such as cell differentiation, cell polarity, cell adhesion and migration [4]. Combined with their often tissue- or cell-specific expression profile, the members of the adhesion GPCR family form a whole new group of potential drug targets.

In this research, we focused on ADGRG2 (GPR64). ADGRG2 is almost exclusively expressed in the epididymis of the testis, where it has been implied in fluid regulation [6, 7]. Interestingly, the receptor is often overexpressed in carcinomas and Richter et al. [8] reported that ADGRG2 is involved in tumor progression and invasion of Ewing Sarcoma. Knockdown of ADGRG2 impaired growth of a Ewing Sarcoma cell line in vitro, showing a significant decrease in colony formation in a methylcellulose assay and suppressing local tumor growth and lung metastasis formation in Rag2^-/-γc-/- mice. The authors additionally revealed that ADGRG2 influenced expression of placental growth factor (PGF) and matrix metalloproteinase 1 (MMP1), both associated with invasive growth.

4.1 ADGRG2 is involved in cell migration and adhesion, but not proliferation

To gain more insight in the involvement of ADGRG2 in cell migration, adhesion and proliferation we used the breast cancer cell line Hs578T as a model system. ADGRG2 is expressed both in Hs578T and MDA-MB-231 cell lines [8] (and data not shown). By means of a smart pool of ADGRG2 siRNA, consisting of four individual siRNAs, we transiently knocked down ADGRG2 expression and monitored the effect on cell behavior. A phagokinetic track assay revealed that migratory tracks of ADGRG2 knockdown cells cover a much shorter distance and form significantly less protrusions compared to control cells. Live monitoring of the ADGRG2 knockdown cells confirmed this phenotype and showed that migratory speed was strongly reduced too. Investigation of the four single siRNAs individually proved that the phenotype was not caused by off-target effects of one of the siRNAs within the smart pool.

Besides migration, also cell adhesion and spreading proved to be affected by transient knockdown of ADGRG2 as shown by measuring adhesion-induced impedance and live imaging of Hs578T cells in real time. The curves obtained with the impedance based xCELLigence technique also allowed us to distinguish between the adhesion and proliferation phase. Where the adhesion

(18)

41 phase was markedly delayed, no change in the proliferation phase was observed. This additionally proves that ADGRG2 knockdown does not affect cell viability.

Comparable results were obtained in another breast cancer cell line, MDA-MB-231, indicating that the effect is not specific for Hs578T and can probably be extended to other carcinoma derived cell lines in which ADGRG2 is upregulated.

4.2 ADGRG2 constitutively activates RhoA-SRE and Gαq –NFκB pathways

GPCRs have been associated with cell migration and adhesion via activation of heterotrimeric G proteins, G protein independent pathways and even via transactivation of tyrosine kinase receptors and subsequent activation of adhesion and migration related transcription factors [35].

For example, activation of both Gαs and Gαq and its downstream signaling events can result in NFκB activation [36], regulating genes controlling apoptosis, cell adhesion, (tumor) cell migration and invasion [37]. RhoA and subsequent ROCK activation by Gα12/13 has been shown to lead to SRE dependent gene transcription mediating cell contraction and migration [38].

We found that ADGRG2 is able to constitutively activate both NFκB and SRE. Moreover, application of small molecule pathway inhibitors indicated that part of the activation NFκB originates from Gαq protein activation and SRE activity from activated RhoA potentially via Gα12/13

protein coupling. This is the very first report that ADGRG2 is a functional receptor, able to activate intracellular signaling pathways that originate from G protein coupling.

GPCRs often couple to multiple G proteins. In fact, so far all known Gα12/13-linked GPCRs additionally signal through another Gα subunit like Gαi/o and/or Gαq/11 [39]. Gαq activating receptors often have the ability to additionally promote cell migration through Gα12/13 [35]. One example is the Sphingosine-1-phosphate receptor 2 (S1P2) that influences chemotaxis via both Gαq and Gα12/13 in vascular smooth muscle cells [40].

4.3 ADGRG2 displays constitutive biased signaling

Interestingly, while ADGRG2 is able to constitutively activate both the SRE and NFκB pathways, it appears to do so differently. This was already apparent when we measured SRE and NFκB activity of several truncated constructs of human ADGRG2 in HEK293 cells, where the receptor lacking the N-terminal 338 amino acids (ADGRG2-GAIN) was able to constitutively activate the SRE, but not the NFκB pathway. We subsequently investigated the signaling properties of a cleavage deficient mutant, T608A. The equivalent position in the ADGRL1 (LPHN1) was previously shown to be the exact site of autoproteolysis and mutating the position resulted in an uncleaved receptor [16]. We noticed that constitutive activity of the ADGRG2-T607A mutant in the NFκB pathway was unchanged, while SRE activation was significantly decreased. We therefore conclude that the conformational change induced by autoproteolytic cleavage is essential for activation of SRE and that the NFκB pathway is not dependent on this particular conformation.

Not many examples exist of this kind of biased constitutive activity. However, in the class A GPCR

(19)

Chapter 3

42

CC-chemokine receptor 5 (CCR5) a mutation in CCR5 at the TM6/7 interface altered the receptor conformation in such a way that Gαi constitutive activity increased, while basal β-arrestin recruitment decreased [41]. It is not unlikely that other GPCRs, especially those that can be modified by for instance proteolysis, could display a similar phenomenon.

4.4 The N-terminus of ADGRG2 acts as positive regulator

Adhesion GPCRs are characterized by the presence of a GPCR-Autoproteolysis Inducing Domain (GAIN domain) of approximately 320 amino acids that includes the GPCR Proteolysis Site (GPS) where the actual cleavage motif is found. The GAIN domain is composed of six α-helices followed by a twisted β-sandwich that includes 13 β-strands and 2 small α-helices. The whole domain is necessary for autoproteolysis that already occurs during protein biosynthesis [5]. The two adhesion GPCR protein domains do stay in close proximity of each other and will be incorporated in the cell membrane as a non-covalently bound complex where it would be able to bind extracellular ligands and subsequently activate intracellular signaling pathways (reviewed by Langenhan et al. [14]). Little is known about the presence and origin of such ligands and the specific activation mechanisms involved. However, for a number of adhesion GPCRs it has been reported that deletion of the complete N-terminal domain, expressing only the 7TM domain from the cleavage site, leads to an increase in constitutive activity [18-22]. This led to the consensus that a general activation mechanism exists among adhesion GPCRs, where the N-terminus functions as an intrinsic inhibitor of receptor activation and that the N-terminus would have to physically dissociate from the 7TM domain to allow transmission of the full biological signal.

Here we report that at least one adhesion GPCR is an exception to the rule and that the N- terminus of ADGRG2 is in fact acting as a positive regulator, maintaining a conformational equilibrium that allows for constitutive activity of the receptor. Removing the N-terminus disrupts all activation of SRE and NFκB. This does not mean that no endogenous agonists exist that can activate the receptor even further. However, it is not likely that evoked dissociation of the N-terminus is part of the ligand-activating mechanism as has been suggested for other adhesion GPCRs [4, 15]. Even though many ADGRs are being activated through similar activation mechanisms, it is likely that some of them are activated in a different manner. For instance, ADGRA1 (GPR123) is the only adhesion GPCR that does not contain a GAIN domain [42] and ADGRF2 (GPR111) and ADGRF4 (GPR115) proved not to undergo autoproteolysis in HEK293 cells [43].

4.5 ADGRG2 is involved in the non-canonical NFκB/RelB pathway in Hs578T cells

In order to gain insight whether SRE and NFκB activation by ADGRG2 could be responsible for cell migration and adhesion, we performed a western blot analysis of several pathway components.

Unfortunately, we were not able to observe any changes on ERK phosphorylation upon ADGRG2 knockdown down in Hs578T cells. Therefore, we cannot confirm an involvement of the RhoA-

(20)

43 ERK-SRE pathway in Hs578T cell adhesion and migration. Whole cell constitutive activation of ERK appears to be quite high in WT Hs578T cells, even under serum deprived conditions, that might conceal the loss of ADGRG2-related ERK phosphorylation after knockdown. Interestingly, knockdown of SRE signaling pathway components in a larger siRNA screen, including RhoA, Gα12

and Gα13, also resulted in reduced migration and low axial ratio in a PKT assay (M. Fokkelman and V.M. Rogkoti, data unpublished).

The western blot analysis did reveal an effect on the NFκB pathway upon ADGRG2 knockdown.

Here, we observed a downregulation of the NFκB family member RelB, but not RelA. RelB is involved in the non-canonical NFκB pathway activated by e.g. lipopolysaccharides, lymphotoxin receptors and latent membrane protein-1 [37]. Thus far, no other GPCR has been described to activate this particular route. Previous studies have identified the non-canonical NFκB pathway to be involved in breast cancer cell migration. For example, constitutive RelB activation proved to be increased in oestrogen receptor alpha (ERα)-negative breast cancer cells and siRNA knockdown of RelB in various breast cancer cells including Hs578T reduced cell migration and invasion [44]. Van Roosmalen et al. performed a siRNA PKT screen of ca. 1500 genes encoding kinases and migration-related proteins. The authors show that the NFκB pathway was most affected and involved the non-canonical pathway components MAP3K14 (NIK) and RelB [31]. Our data now points to ADGRG2 being responsible for at least part of the observed constitutive RelB activation.

(21)

Chapter 3

44

5. Conclusions

We summarized our findings in a schematic model depicted in Figure 7. Here, ADGRG2 constitutively activates the (Gα12/13)-RhoA-SRE and the Gαq-NFκB pathways. Of the two signaling pathways, only SRE signaling is dependent on autoproteolysis. Our data suggests that a non- canonical NFκB pathway is activated downstream of Gαq leading to RelB activation and subsequent cell adhesion and migration of Hs578T cells. We were unable to confirm whether the constitutive SRE activation of ADGRG2 is involved in this process as well. We can also not exclude any direct role of the large N-terminal domain of ADGRG2 in cell adhesion. Most adhesion GPCR family members possess large extracellular N-termini that contain protein domains similar to motifs involved in cell adhesion. The domains have been indicated in some family members to interact with extracellular matrix proteins or are the site for endogenous ligand binding, e.g.

ADGRE5 (CD97) and ADGRG1 (GPR56) [45, 46]. It is probable that one or more similar N-terminal binding domains (NBD) are present in ADGRG2, although no such domain has yet been described.

By establishing the involvement of ADGRG2 in cell adhesion and migration and the presence of classical GPCR signaling properties, we have made a first step in characterizing ADGRG2 as a potential drug target for the treatment of metastasizing carcinomas. It would be highly interesting to screen for agents that are able to disrupt the N-terminal-7TM interaction or that are directly able to act as inverse agonists of the receptor’s constitutive activity, hereby reducing the signal transduction of the receptor and limiting cancer cell adhesion and migration.

(22)

45 Figure 7. Schematic model of ADGRG2 activation and its role in adhesion and migration

ADGRG2 constitutively activates a non-canonical NFκB pathway downstream of Gαq leading to RelB activation and subsequent cell adhesion and migration of Hs578T cells (right solid arrow). ADGRG2 also constitutively activates the RhoA-SRE pathway that is dependent on receptor autocleavage and likely originates from Gα12/13 coupling (left dashed arrow).

Next to intracellular signaling pathways, a putative N-terminal Binding Domain (NBD) located on the extracellular domain of ADGRG2 might also be directly involved in cell adhesion. A ligand or binding partner has not yet been identified for ADGRG2.

(23)

Chapter 3

46

References

[1] Foord SM, Bonner TI, Neubig RR, Rosser EM, Pin JP, Davenport AP, Spedding M, Harmar AJ, Pharmacol Rev.

2005;57:279-288.

[2] Paavola KJ, Hall RA, Molecular pharmacology. 2012;82:777-783.

[3] Osterhoff C, Ivell R, Kirchhoff C, DNA and cell biology. 1997;16:379-389.

[4] Hamann J, Aust G, Arac D, Engel FB, Formstone C, Fredriksson R, Hall RA, Harty BL, Kirchhoff C, Knapp B, Krishnan A, Liebscher I, Lin HH, Martinelli DC, Monk KR, Peeters MC, Piao X, Promel S, Schoneberg T, Schwartz TW, Singer K, Stacey M, Ushkaryov YA, Vallon M, Wolfrum U, Wright MW, Xu L, Langenhan T, Schioth HB, Pharmacological reviews.

2015;67:338-367.

[5] Arac D, Boucard AA, Bolliger MF, Nguyen J, Soltis SM, Sudhof TC, Brunger AT, The EMBO journal. 2012;31:1364- 1378.

[6] Kirchhoff C, Osterhoff C, Samalecos A, Reproduction. 2008;136:235-245.

[7] Davies B, Baumann C, Kirchhoff C, Ivell R, Nubbemeyer R, Habenicht UF, Theuring F, Gottwald U, Molecular and cellular biology. 2004;24:8642-8648.

[8] Richter GH, Fasan A, Hauer K, Grunewald TG, Berns C, Rossler S, Naumann I, Staege MS, Fulda S, Esposito I, Burdach S, The Journal of pathology. 2013;230:70-81.

[9] Koirala S, Jin Z, Piao X, Corfas G, The Journal of neuroscience : the official journal of the Society for Neuroscience.

2009;29:7439-7449.

[10] Saito Y, Kaneda K, Suekane A, Ichihara E, Nakahata S, Yamakawa N, Nagai K, Mizuno N, Kogawa K, Miura I, Itoh H, Morishita K, Leukemia. 2013;27:1637-1649.

[11] Iguchi T, Sakata K, Yoshizaki K, Tago K, Mizuno N, Itoh H, The Journal of biological chemistry. 2008;283:14469- 14478.

[12] Tang X, Jin R, Qu G, Wang X, Li Z, Yuan Z, Zhao C, Siwko S, Shi T, Wang P, Xiao J, Liu M, Luo J, Cancer research.

2013;73:6206-6218.

[13] Galle J, Sittig D, Hanisch I, Wobus M, Wandel E, Loeffler M, Aust G, The American journal of pathology.

2006;169:1802-1811.

[14] Langenhan T, Aust G, Hamann J, Science signaling. 2013;6:re3.

[15] Liebscher I, Ackley B, Arac D, Ariestanti DM, Aust G, Bae BI, Bista BR, Bridges JP, Duman JG, Engel FB, Giera S, Goffinet AM, Hall RA, Hamann J, Hartmann N, Lin HH, Liu M, Luo R, Mogha A, Monk KR, Peeters MC, Promel S, Ressl S, Schioth HB, Sigoillot SM, Song H, Talbot WS, Tall GG, White JP, Wolfrum U, Xu L, Piao X, Annals of the New York Academy of Sciences. 2014;1333:43-64.

[16] Krasnoperov V, Lu Y, Buryanovsky L, Neubert TA, Ichtchenko K, Petrenko AG, The Journal of biological chemistry.

2002;277:46518-46526.

[17] Arac D, Aust G, Calebiro D, Engel FB, Formstone C, Goffinet A, Hamann J, Kittel RJ, Liebscher I, Lin HH, Monk KR, Petrenko A, Piao X, Promel S, Schioth HB, Schwartz TW, Stacey M, Ushkaryov YA, Wobus M, Wolfrum U, Xu L, Langenhan T, Annals of the New York Academy of Sciences. 2012;1276:1-25.

[18] Liebscher I, Schon J, Petersen SC, Fischer L, Auerbach N, Demberg LM, Mogha A, Coster M, Simon KU, Rothemund S, Monk KR, Schoneberg T, Cell reports. 2014;9:2018-2026.

[19] Stoveken HM, Hajduczok AG, Xu L, Tall GG, Proceedings of the National Academy of Sciences of the United States of America. 2015.

[20] Okajima D, Kudo G, Yokota H, Journal of receptor and signal transduction research. 2010;30:143-153.

[21] Paavola KJ, Stephenson JR, Ritter SL, Alter SP, Hall RA, The Journal of biological chemistry. 2011;286:28914- 28921.

[22] Ward Y, Lake R, Yin JJ, Heger CD, Raffeld M, Goldsmith PK, Merino M, Kelly K, Cancer research. 2011;71:7301- 7311.

[23] Naffar-Abu-Amara S, Shay T, Galun M, Cohen N, Isakoff SJ, Kam Z, Geiger B, PloS one. 2008;3:e1457.

[24] Ezratty EJ, Partridge MA, Gundersen GG, Nature cell biology. 2005;7:581-590.

[25] Carpenter AE, Jones TR, Lamprecht MR, Clarke C, Kang IH, Friman O, Guertin DA, Chang JH, Lindquist RA, Moffat J, Golland P, Sabatini DM, Genome biology. 2006;7:R100.

[26] Yan K, Verbeek F, in: Margaria T, Steffen B, (Eds.), Leveraging Applications of Formal Methods, Verification and Validation. Applications and Case Studies. Springer Berlin Heidelberg2012. pp. 25-41.

(24)

47 [27] Wink S, Hiemstra S, Huppelschoten S, Danen E, Niemeijer M, Hendriks G, Vrieling H, Herpers B, van de Water B, Chemical research in toxicology. 2014;27:338-355.

[28] Hodges RS, Heaton RJ, Parker JM, Molday L, Molday RS, The Journal of biological chemistry. 1988;263:11768- 11775.

[29] Guan XM, Kobilka TS, Kobilka BK, The Journal of biological chemistry. 1992;267:21995-21998.

[30] Liu H, Naismith JH, BMC biotechnology. 2008;8:91.

[31] van Roosmalen W, Le Devedec SE, Golani O, Smid M, Pulyakhina I, Timmermans AM, Look MP, Zi D, Pont C, de Graauw M, Naffar-Abu-Amara S, Kirsanova C, Rustici G, Hoen PA, Martens JW, Foekens JA, Geiger B, van de Water B, The Journal of clinical investigation. 2015;125:1648-1664.

[32] Krishnan A, Almen MS, Fredriksson R, Schioth HB, PloS one. 2012;7:e29817.

[33] Nordstrom KJ, Lagerstrom MC, Waller LM, Fredriksson R, Schioth HB, Molecular biology and evolution.

2009;26:71-84.

[34] Bjarnadottir TK, Fredriksson R, Schioth HB, Cellular and molecular life sciences : CMLS. 2007;64:2104-2119.

[35] Cotton M, Claing A, Cellular signalling. 2009;21:1045-1053.

[36] Fraser CC, International reviews of immunology. 2008;27:320-350.

[37] Perkins ND, Nature reviews. Molecular cell biology. 2007;8:49-62.

[38] Siehler S, British journal of pharmacology. 2009;158:41-49.

[39] Riobo NA, Manning DR, Trends in pharmacological sciences. 2005;26:146-154.

[40] Takashima S, Sugimoto N, Takuwa N, Okamoto Y, Yoshioka K, Takamura M, Takata S, Kaneko S, Takuwa Y, Cardiovascular research. 2008;79:689-697.

[41] Steen A, Thiele S, Guo D, Hansen LS, Frimurer TM, Rosenkilde MM, The Journal of biological chemistry.

2013;288:12511-12521.

[42] Fredriksson R, Gloriam DE, Hoglund PJ, Lagerstrom MC, Schioth HB, Biochemical and biophysical research communications. 2003;301:725-734.

[43] Promel S, Waller-Evans H, Dixon J, Zahn D, Colledge WH, Doran J, Carlton MB, Grosse J, Schoneberg T, Russ AP, Langenhan T, Developmental dynamics : an official publication of the American Association of Anatomists.

2012;241:1591-1602.

[44] Wang X, Belguise K, Kersual N, Kirsch KH, Mineva ND, Galtier F, Chalbos D, Sonenshein GE, Nature cell biology.

2007;9:470-478.

[45] Hamann J, Vogel B, van Schijndel GM, van Lier RA, The Journal of experimental medicine. 1996;184:1185-1189.

[46] Luo R, Jeong SJ, Jin Z, Strokes N, Li S, Piao X, Proceedings of the National Academy of Sciences of the United States of America. 2011;108:12925-12930.

(25)

Chapter 3

48

Acknowledgements

This work was supported by grants from the EU-FP7 - Systems Microscopy NoE (grant no. 258068 to B. van de Water).

Supplementary Data

Supplementary movies to this article can be found online at:

http://dx.doi.org/10.1016/j.cellsig.2015.08.015.

Supplementary Movie S1: siRNA knockdown of ADGRG2 in Hs578T reduces single cell migration speed. Hs578T-GFP cells were transfected with ADGRG2 smartpool siRNA. After 65h, knockdown and control cells were harvested and replated on fibronectin-coated microscopy plates. Cells were allowed to adhere for 10h before imaging started. 2D migration of Hs578T mock and ADGRG2 knockdown cells was followed by live cell microscopy over a period of 12 h.

Supplementary Figure S2: siRNA mediated knockdown in MDA-MB-231 confirms the role of ADGRG2 in cell migration. To confirm the reduced migration of ADGRG2 knockdown, MDA-MB- 231-GFP cells were transfected as described for Hs578T cells. Cell migration was monitored by live cell microscopy over a period of 12h, after which individual cells were tracked and single cell speed was quantified. Both smartpool (si-ADGRG2 sp) and all single siRNA sequences (#1-4) reduce migration speed compared to MDA-MB-231 mock cells. Graph shows single cell populations with mean ± 95% confidence interval and statistical significance was determined using Kruskal-Wallis test with Dunn’s post correction (*** = p<0.001, **** = p<0.0001). Data shown represents one replicate of three independent experiments.

MDA-MB-231 live cell migration

migrationspeed(um/h)

mock si-DNM2

si-ADGRG2sp si-ADGRG2#1

si-ADGRG2#2 si-ADGRG2#3

si-ADGRG2#4 0

20 40

60 **** **** **** *** **** ****

(26)

49 Supplementary Movie S3: ADGRG2 knockdown affects Hs578T adhesion. siRNA transfected Hs578T cells were seeded on fibronectin-coated microscopy plates and cell adhesion was captured directly using DIC microscopy. Cell adhesion is delayed after knockdown of ADGRG2 compared to mock.

Supplementary Figure S4: MDA-MB-231 cells show delayed adhesion after ADGRG2 knockdown. MDA-MB-231 cells start to adhere after approx. 20 minutes, whereas ADGRG2 knockdown cells require 40 – 50 minutes to start spreading.

(27)

50