Reduced pathological angiogenesis and tumor growth ... - Springer Link

3 downloads 0 Views 763KB Size Report
Nov 2, 2011 - ORIGINAL PAPER. Reduced pathological angiogenesis and tumor growth in mice lacking GPR4, a proton sensing receptor. Lorenza Wyder ...
Angiogenesis (2011) 14:533–544 DOI 10.1007/s10456-011-9238-9

ORIGINAL PAPER

Reduced pathological angiogenesis and tumor growth in mice lacking GPR4, a proton sensing receptor Lorenza Wyder • Thomas Suply • Be´range`re Ricoux • Eric Billy • Christian Schnell • Birgit U. Baumgarten • Sauveur Michel Maira • Claudia Koelbing • Mireille Ferretti • Bernd Kinzel • Matthias Mu¨ller Klaus Seuwen • Marie-Gabrielle Ludwig



Received: 20 June 2011 / Accepted: 11 October 2011 / Published online: 2 November 2011 Ó Springer Science+Business Media B.V. 2011

Abstract The G protein-coupled receptor GPR4 is activated by acidic pH and recent evidence indicates that it is expressed in endothelial cells. In agreement with these reports, we observe a high correlation of GPR4 mRNA expression with endothelial marker genes, and we confirm expression and acidic pH dependent function of GPR4 in primary human vascular endothelial cells. GPR4-deficient mice were generated; these are viable and fertile and show no gross abnormalities. However, these animals show a significantly reduced angiogenic response to VEGF (vascular endothelial growth factor), but not to bFGF (basic fibroblast growth factor), in a growth factor implant model. Accordingly, in two different orthotopic models, tumor growth is strongly reduced in mice lacking GPR4. Histological analysis of tumors indicates reduced tumor cell proliferation as well as altered vessel morphology, length and density. Moreover, GPR4 deficiency results in reduced VEGFR2 (VEGF Receptor 2) levels in endothelial cells, accounting, at least in part, for the observed phenotype. Our data suggest that endothelial cells sense local tissue acidosis via GPR4 and that this signal is required to generate a full angiogenic response to VEGF.

Electronic supplementary material The online version of this article (doi:10.1007/s10456-011-9238-9) contains supplementary material, which is available to authorized users. L. Wyder  T. Suply  B. Ricoux  E. Billy  C. Schnell  B. U. Baumgarten  S. M. Maira  C. Koelbing  M. Ferretti  B. Kinzel  M. Mu¨ller  K. Seuwen  M.-G. Ludwig (&) Novartis Institutes for Biomedical Research, Forum1 Novartis Campus, CH-4056 Basel, Switzerland e-mail: [email protected] Present Address: L. Wyder Actelion Pharmaceuticals Ltd, Allschwil, Switzerland

Keywords Angiogenesis  Acidosis  Hypoxia  Endothelium  Tumor

Introduction GPR4 belongs to a protein family comprising 3 closely related G protein-coupled receptors (GPCRs): GPR4, OGR1/GPR68 and TDAG8/GPR65. We have previously shown that OGR1 as well as GPR4 sense extracellular protons and stimulate intracellular second messengers upon exposure to slightly acidic pH [1]. Similarly, TDAG8 has been identified as a proton-sensing receptor [2, 3]. Half maximal activation of these receptors is observed in the physiological range, around pH 7.4, and highest activity is observed at pH 6.8. GPR4 and OGR1 had originally been reported as receptors for lipid messengers, however, these publications have recently been withdrawn [reviewed in 4]. In gene expression profiling studies, we found a strong correlation between the expression of GPR4 mRNA and marker genes for endothelial cells. Earlier reports had already described an important role for GPR4 in endothelial cell function, but these findings were seen in relation to the presumed ligands sphingosylphosphorylcholine and lysophosphatidylcholine [5–9]. The first study reporting the phenotype of mice lacking GPR4 appeared recently and confirmed a role of this receptor in vascular biology [10]. While adult GPR4-deficient (GPR4 KO) mice appeared normal, some embryos and neonates had hemorrhages and abnormal blood vessels, explaining reduced litter sizes for GPR4 KO intercrosses. In in vitro aortic ring assays, microvessel outgrowth was less dependent on medium pH when aortas from GPR4 KO were used compared to wild type [10].

123

534

Angiogenesis, the formation of new blood vessels, is a hallmark of cancer, allowing tumors to grow beyond 1–3 mm3 in size and facilitating local invasion and metastasis. It is induced by aberrant expression of angiogenic growth factors such as VEGF but also by local alteration of the tumor microenvironment through hypoxia, glucose deprivation, and oxidative and mechanical stress [11]. Recently, several anti-VEGF and anti-VEGF-receptor therapies have been shown to reduce vascularization and tumor growth, validating the notion that angiogenesis is a relevant target for cancer therapy [12]. As a response to hypoxia, tumor cells increase their glycolytic rate to produce energy and thereby acidify the extracellular space [13–15]. Upregulated glycolysis can indeed be observed in most tumors by FdG (fluorodesoxyglucose) PET (positron-emission tomography), a commonly used imaging method to diagnose tumors [16, 17]. By adapting to hypoxia and acidosis, tumor cells survive under conditions that are not tolerated by normal cells and this feature may correlate with invasive potential and resistance to chemotherapy [18–20]. Mechanisms by which hypoxia-induced HIF-1a hypoxia-inducible factor) modulates intracellular pH are beginning to be unraveled [15, 21]. However, little is known about how cells can sense and adapt to extracellular acidosis and the interplay between acidosis and angiogenesis is not yet fully understood. To gain a better understanding of the role of GPR4 in physiological and pathological angiogenesis, we have generated GPR4-deficient mice, independently of the study reported by Yang et al. [10]. We report that adult animals are viable and fertile and do not show major abnormalities. However, the GPR4-deficient mice show strongly reduced responses to VEGF-driven, but not bFGF-driven, angiogenesis when subjected to a growth factor implant angiogenesis model. In addition, tumor growth is reduced in GPR4-deficient mice compared to wild-type mice in two different orthotopic tumor models. Reduced tumor growth correlates with impaired vessel structure as well as reduced VEGFR2 levels in GPR4-deficient mice. Our data suggest that acidosis is sensed by endothelial cells via GPR4, and that this signal can modulate pathological angiogenesis.

Angiogenesis (2011) 14:533–544

Using 133A Affymetrix oligonucleotide array technology 109 chips (cell lines, primary cells and various human tissues) were selected where the Flag was present (41 chips) and absent (68 chips), respectively, for most or all of these markers. The MAS5 normalized expression data of the selected chips were analysed with GeneSpring (gene expression software provided by Silicon GeneticsÒ). For each of the 7 marker genes all genes with a similar expression profile were determined using the Pearson correlation. Only genes with a correlation coefficient of 0.7 or above to at least one of the 7 marker genes were considered. RNA extraction and RT-PCR Total RNA was recovered from cells using the QIAGEN RNA-easy Mini kit and following the manufacturer’s instructions. Prior to reverse transcription, RNAs were treated with DNase I according to manufacturer’s procedure (Amplification Grade DNaseI, Invitrogen, Basel, Switzerland). Reverse transcription was performed in a total volume of 30 ll. The reaction mix contained: 20 ll of 2 lg DNaseI-treated RNA, 2 ll of 109 Buffer RT, 2 ll dNTP mix (5 mM each dNTP), 2 ll oligo-dT primer (10 lM), 1 ll RNAse inhibitor (10 U/ll), 1 ll Omniscript reverse transcriptase (QIAGEN, Basel, Switzerland), 2 ll RNAse-free water. The mixture was incubated for 60 min at 37°C, then the reaction was heat-inactivated at 93°C for 5 min and then rapidly cooled down on ice. PCR was performed with 5 ll of cDNA and using the Hotstart Mastermix kit (QIAGEN, Basel, Switzerland). Amplification was with the following program: initial denaturation at 95°C for 15 min followed by 45 cycles of 15 s at 94°C, 30 s at 50°C and 30 s at 72°C. Primers used were: mouse GPR4-1466F (TGTGCTACCGTGGCATC CT), Mouse GPR4-1581R (AAAGCACACCAGCACAAT GG), mouse GAPDH-F960 (TTGTCAAG CTCATTTCCT GGTATG) mouse GAPDH-1062R (TGGTCCAGGGTTT CTTACTCCTT); human GPR4-2319F (TGTGCTACCGT GGCATCCT), human GPR4-2469R (CTTGAGTTC TGA CATTCTCCCTCTT); human GAPDH-111F (CAGGGCT GCTTTTAACTCTGGTA); human GAPDH-211R (GGGT GGAATCATATTGGAACATG). Generation of stable HeLa-GPR4 cells

Materials and methods Bioinformatics screen Seven marker genes (official human gene symbols: KDR, TIE1, TEK, ANGPT2, CDH5, VWF, PTPRB) known to be expressed almost exclusively in endothelial cells and to show similar expression profiles were selected as reference.

123

For the generation of the stable HeLa GPR4 cell line the pcDNA3.1(?)/myc-His vector (Invitrogen, Basel, Switzerland) expressing full length human GPR4 was linearized with PvuI and transfected using Effectene reagent (QIAGEN, Basel, Switzerland). Stable cell clones expressing the receptor were isolated following selection in the culture medium at pH 7.4, without Hepes and with antibiotic G418

Angiogenesis (2011) 14:533–544

(400 lg/ml). After 18 days, cells were grown further at pH 7.9. Cell culture Hela-GPR4 stable cells were grown in a 1:1 mixture of bicarbonate-buffered DMEM and Ham’s F12 medium supplemented with 10 mM Hepes, 10% fetal calf serum and antibiotics at pH 7.9. HUVEC cells were purchased from Promo Cell (BioConcept AG, Allschwil, Switzerland, C-10251), and cultured in Medium C-22210 plus Supplement kit C-39210 (both from Promocell/BioConcept AG, Allschwil, Switzerland) and a final concentration of 5% fetal calf serum (South American10270-106, Gibco/Invitrogen, Basel, Switzerland). 4T1 mouse breast cancer cell lines were obtained from ATCC (LGC Promochem, Molsheim, France) and grown in DMEM high glucose, 1% Glutamine, 10% FCS. CT26 mouse colon carcinoma cells were obtained from ATCC and grown in MEM, 10% FCS,1% sodium pyruvate,1% Glutamine, 1% non-essential amino acids, 2% vitamins. cAMP formation assay Confluent cell cultures grown in 24-well plates were labeled with [3H]adenine (100 MBq/ml; Amersham, Zuerich, Switzerland) for 4 h in serum-free DMEM medium. Cells were then incubated at 37°C in buffered salt solution containing 130 mM NaCl, 0.9 mM NaH2PO4, 5.4 mM KCl, 0.8 mM MgSO4, 1.0 mM CaCl2, 25 mM glucose, 20 mM Hepes. The pH of all solutions was adjusted to the indicated value at room temperature. The phosphodiesterase inhibitor isobutylmethylxanthine (IBMX, 1 mM), was added as indicated to allow accumulation of cAMP. Forskolin (FSK) activates adenylyl cyclases in synergy with Gas stimulation and was therefore used to increase the assay window. Incubation time was 15 min. Cells were then extracted with ice-cold trichloroacetic acid and cAMP separated from free adenine and ATP using batch column chromatography according to the method described by Salomon [22]. siRNA All siRNAs were designed using a proprietary algorithm (Novartis Nucleic Acid Science unit, Basel, Switzerland), and were synthesized by QIAGEN. A standardized mRNA fusion-construct assay was used to screen several different siRNAs for their potency in targeting human and mouse GPR4, respectively [23]. The most potent siRNAs were used in this study. Lyophilized siRNAs were resuspended

535

in the provided hybridization buffer prior usage. As a control siRNA (siCtrl) the non-targeted siRNA from QIAGEN was used. HUVEC cells (passage 3) were transfected with Hiperfect (QIAGEN) according to manufacturer’s instructions, 3 ll of Hiperfect transfection reagent were used for 30,000 cells in a 24-well and the final siRNA concentration was 10nM. 48 h after transfection cells were harvested for RNA purification, FACS analysis or used in the cAMP formation assay. Sequences: siMm1-GPR4: 50 -GGGTCTGAAGGGGGAACAAdTdT-30 siMm2-GPR4: 50 -GGAGGTAGGACTAACAATAdTdT-30 siRNA HuGPR4: 50 -CTGCGCTGTGTCCTATCTCAA-30 siRNA VEGFR2: 50 -ATGAAAGTTACCAGTCTATTA-30 Qiagen Control no target siRNA: 50 -AATTCTCCGAA CGTGTCACGT-30 Animals Female Balb/C mice (WT or GPR4 KO) of 68 weeks of age were bred at the Novartis animal breeding facility. Control BALB/c mice were obtained from Charles River Laboratories (Les Oncins, France). Mice had unrestricted access to food and water and all procedures were carried out in accordance with the Swiss law for animal protection. Mice were identified by ear markings and kept in groups (5–6 animals per cage) under normal conditions and observed daily. Five to ten mice were used per treatment group. Experiments were performed at least twice. Studies concerning the angiogenesis and tumor models described in this report were performed according to animal experimentation licenses no. 1325 and 1402, approved by the Basel Stadt Authorities. Angiogenesis growth factor implant model The model has been previously described [24–26]. Briefly, porous tissue chambers made of perfluoro-alkoxy-Teflon (TeflonÒ-PFA, 21 9 8 mm diameter, 550 ll volume) were filled with 0.8% agar (BBLÒ Nr. 11849, Becton–Dickinson, Meylan, France) and 20 U/ml heparin, (Roche, Basel, Switzerland) supplemented with or without 3 lg/ml recombinant human VEGF165 and 0.3 mg/ml bFGF (Invitrogen, Basel, Switzerland) and siRNAs as indicated. Solutions were maintained at 39°C prior to the filling procedure. Mice were anesthetized using 3% Isoflurane (ForeneÒ, Abbott AG, Cham, Switzerland) inhalation. For subcutaneous implantation, a small skin incision was made at the base of the tail to allow the insertion of an implant trocar. The chamber was implanted under aseptic conditions through the small incision onto the back of the

123

536

animal. The skin incision was closed by wound clips (Autoclip 9 mm Clay Adams). On the 4th day after implantation, animals were sacrificed using CO2. For the siRNA experiments, siRNAs were added at a final concentration of 0.3 lM together with the growth factors into the chambers and animals were sacrificed 3 days after implantation. Chambers were excised and the vascularized fibrous tissue formed around each implant carefully removed and weighed. Body weight was used to monitor the general condition of the mice. Quantification of the angiogenic response The fibrous tissue that grew around the implant was homogenized for 30 s at 24,000 rpm (Ultra Turrax T25) after addition of 1 ml RIPA buffer (50 mM Tris–HCl pH7.2, 120 mM NaCl, 1 mM EDTA pH8.0, 6 mM EGTA pH8.5, 1% NP-40, 20 mM NaF) to which 1 mM Pefabloc SC Proteinase inhibitor (Roche, Basel, Switzerland) and 1 mM Na-Vanadate were freshly added. The homogenate was centrifuged for 30 min at 7,000 rpm and the supernatant was filtered using a 0.45 lm GHP syringe filter (AcrodiscÒ GF, Gelman Sciences, Ann Arbor, MI) to avoid fat contamination. This lysate was used for measuring Tie2 protein levels by ELISA as described [27]. Orthotopic tumor models Twenty microliter of 4T1 cell suspension (5 9 107 cells/ml in PBS) were injected under light isoflurane anesthesia, under the fat pad of the 4th mammary gland, to give a total inoculum of 106 cells per mouse. For the CT26 tumor model, cell inoculation was performed in an OHC zone, under aseptic conditions. Prior to surgery, the mice were anaesthetized with a single subcutaneous injection of a freshly prepared mixture of ketamine (100 mg/kg) and xylazine (5 mg/kg). The abdomen (skin and muscle) was opened with surgical scissors along the linea alba (0.5–1 cm long). With neutral forceps, the caecum of the animal was located, and gently pulled out from the abdomen. With a 30 Gauge needle (Becton–Dickinson, 320834), 10 ll of a CT26 cell suspension (106 cells/ml in HBSS) was injected in the submucosa layer with the assistance of a magnifying glass, to give a total inoculum of 1 9 105 cells per mouse. Finally, the caecum was placed back into the abdomen, the muscle was sutured (Dexon, 9104-11), and the wound was closed with 3 AutoclipsÒ (Clay-Adams 427631). Animals were finally transferred to a 37°C heating pad to recover from the anesthesia. For experiments using PTK787/ZK222584 (PTK787, [25]), the compound, synthesized at Novartis Institutes for Biomedical Research, was administered in a vehicle of 100%

123

Angiogenesis (2011) 14:533–544

PEG-300, from day 7 to 21 p.o., once a day at 100 mg/kg. For both models, animals were sacrificed 19–22 days after tumor implantation. Body weight was used to monitor the general condition of the mice. Immunofluorescence and vessel counts Eight lm cryosections sections of CT26 orthotopic tumors were fixed with 4% formaldehyde, blocked with 10% normal goat serum (NGS) and incubated with primary antibodies diluted 1:200 (desmin 1:100) in PBS/0.5% NGS/0.05% TritonX-100 for 2 h at room temperature. Sections were washed 3 times with PBS and incubated with the secondary antibodies diluted 1:400 in PBS/0.5% NGS/ 0.05% TritonX-100. After 1 h incubation at room temperature, sections were washed with PBS and mounted in Mowiol. Antibodies used were rat anti-mouse CD31 (clone MEC13.3, BD PharMingen, San Diego, CA), rabbit antimouse Ki67 (Neomarkers, Fremont, CA), rabbit anti cleaved Caspase 3 (Asp175; Cell Signalling, Beverly MA), rabbit anti Desmin (clone D93F5, Cell Signalling, Beverly, MA), rabbit anti NG2 (Chemicon/Millipore, Billerica, MA). For smooth muscle actin staining, a FITC conjugated mouse monoclonal anti-alpha smooth muscle actin (SMA) antibody was used (SIGMA, St. Louis, MI)). Secondary antibodies were goat anti-rabbit ALEXA Fluor 568, goat anti-rat ALEXA Fluor 568 and goat anti-rat ALEXA Fluor 488 (all from Molecular Probes, Invitrogen, Basel Switzerland). Ki67 staining was used to quantify the amount of proliferative cells in CT26 tumors. Six to eight representative pictures were taken from each tumor (n = 6 per group) using a Zeiss Axioplan microscope (209 lens). Percent coverage of the total area by proliferating cells was calculated. To determine vessel density, vessels stained for CD31 as described above were counted manually over the whole tumor section. Pictures encompassing the whole tumor where taken at 109 magnification using a Zeiss Axioplan microscope. The area of the counted regions was measured using the Openlab 3.1.5 software (Improvision, Lexington, MA). Six complete tumors were counted per group. As a measure of vascular integrity, the length of continuous CD31-stained structures was determined on micrographs from tumor sections. From each tumor, 4–5 pictures were taken at a 209 magnification using a Zeiss Axioplan microscope. Values (lm) were obtained for about 400 vascular structures per group (wild type n = 420, GPR4KO n = 388). Structures smaller than 7.5 lM were not taken into account. Blood vessel integrity in CT26 tumors was analyzed by injecting mice with tomato lectin-FITC (Vector Laboratories, Burlingame, CA) 3 min before sacrifice.

Angiogenesis (2011) 14:533–544

FACS analysis Non-transfected and siRNA transfected HUVEC cells were analyzed by FACS for VEGFR2 levels. Briefly, cells were trypsinized, washed with PBS ? 10% FCS and incubated 10 min on ice prior to the addition of RPE-conjugated mouse anti human VEGFR2 mAb (1 lg/106 cells; R&D Sytems, Abingdon, UK). RPE-labeled isotype mouse IgG1 was used as FACS control (R&D systems, Abingdon, UK). FACS analysis was performed on a FACS Calibur using Cell Quest Software (Becton–Dickinson, Allschwil, Switzerland). SDS-PAGE, Western-Blot, ELISA Total protein was extracted from tissues with RIPA buffer supplemented with protease inhibitor (Complete, Roche Diagnostics, Switzerland). Proteins were resolved on 8% SDS-PAGE, then blotted onto PVDF membrane and probed with different antibodies (rat anti-mouse VE-Cadherin mAb, clone 11D4.1, Becton–Dickinson, Allschwil, Switzerland; goat anti-mouse EphB4, R&D systems, Abingdon, UK; rabbit anti mouse tubulin, Spring Biosciences, Freemont. CA). Detection was performed with HRPlabeled secondary antibodies and ECL-plus chemiluminescent reagent (Amersham Biosciences, Uppsala, Sweden). Level of Tie2 receptor was determined using a Tie2 ELISA as described [27]. Mouse VEGFR2 and VEGF levels were measured using commercially available ELISA kits (R&D systems, Abingdon, UK). Statistical analysis All data are presented as mean values ± SEM and statistical analysis was performed using the SigmaStat software. For in vivo experiments P-values were calculated with 2-tailed Student’s T-test. In all cases data passed both normality and equal variance tests. P-values \ 0.05 were considered statistically significant. For histological analyses data failed to pass the normality test (P \ 0.001) and thus the Mann–Whitney Rank Sum test was used. P-values \ 0.05 were considered statistically significant.

Results GPR4 acts as a functional proton-sensing receptor in endothelial cells Analysis of gene expression data obtained using DNA microarrays of various cell types and tissues of human origin showed that expression of GPR4 correlated with the

537

expression of a set of marker genes characteristic for endothelial cells (Suppl. Fig. 1). We could confirm specific expression of GPR4 in endothelial cells by extended DNA microarray analysis (Fig. 1a) as well as by RT-PCR in human and mouse endothelial cells (Fig. 1b). To demonstrate that GPR4 is a functional pH sensor on endothelial cells, we exposed HUVECs (primary human umbilical vein endothelial cells) to mild extracellular acidosis and measured cAMP production (Fig. 1c). HeLa cells transfected with recombinant GPR4 were used as controls (Fig. 1d). HUVECs showed a weak but clearly significant response, and this response could be strongly amplified by the addition of forskolin, which activates adenylyl cyclases in synergy with Gas (Fig. 1c). The maximal response was observed around pH 6.8 and there was no significant cAMP production at pH 7.9 in all GPR4 expressing cells, in agreement with our previous results [1]. To demonstrate that GPR4 is the pH sensor responsible for this increased cAMP production, we transfected HUVECs with siRNAs 48 h prior to exposure to extracellular acidosis. A GPR4specific siRNA was able to abrogate the pH-dependent cAMP increase, whereas the control siRNA had no effect (Fig. 1e; Suppl. Fig. 2). GPR4-deficient mice are viable and fertile In order to better understand the function of GPR4, GPR4deficient mice were generated as described in the Online Resource 3 (Suppl. Fig. 3a, b). Expression of GPR4 mRNA was measured by RT-PCR in several organs from both wild type and GPR4-deficient mice. As expected, GPR4 mRNA was absent in all tissues from the GPR4-deficient mice but present in the wild type controls (Suppl. Fig. 3c). GPR4 expression was also shown in primary endothelial cells isolated from lungs of wild type mice, but not in endothelial cells isolated from GPR4-deficient mice (Suppl. Fig. 3d). GPR4-deficient mice are viable and fertile and show no gross abnormalities compared to their wild type littermates. Mendelian ratios of het 9 het litters on BalbC background or fully backcrossed to B6 ([96%) were not significantly affected by GPR4 deficiency, indicating that GPR4 is not essential during development. In addition, no significant histopathological differences were evident in the GPR4deficient mice when compared to age- and gender-matched wild type animals (BalbC background). Notably, the cardiovascular system appeared normal. Impaired VEGF-driven angiogenesis in GPR4-deficient mice To investigate whether GPR4 plays a role in pathological angiogenesis, GPR4-deficient mice were subjected to a

123

538

Angiogenesis (2011) 14:533–544

b

Expression of GPR4 (206236_at)

a 350 300 250 200 150 100 50 0

huGPR4

huGAPDH

mGPR4

Endothelial cells

mGAPDH

Tumor cell lines

Normal cells

2

d

1.5

cAMP formation (dpm x 10-3)

cAMP formation (dpm x 10-3)

c

1

45 40 35 30 25 20 15 10 5 0

n.a.

0.5

IBMX

0

IBMX + FSK

-8

-7.5

-7

-6.5

log [H+] (M)

-8

-7.5

-7

-6.5

12

n.a. - Hela GPR4 IBMX - Hela GPR4 n.a. - Hela IBMX - Hela

10 8 6 4 2 0 -8

-7.5

-7

-6.5

log [H+] (M)

log [H+] (M)

cAMP formation (dpm x 10-3)

e 7 6 5 4 3 2 1 0

pH6.8

7 6 5 4 3 2 1 0

control siRNA

pH7.9

n.a.

IBMX alone

IBMX + FSK

siRNA GPR4-1

n.a.

IBMX alone

IBMX + FSK

Fig. 1 GPR4 is expressed in endothelial cells and functions as a pH sensing receptor. a GPR4 expression in various types of endothelial cells, normal cells and tumor cell lines was determined by microarray experiments. The MAS5 normalized values of GPR4 expression levels are shown (n = 2–3 per sample). b Expression of GPR4 was confirmed by RT-PCR in several human and mouse endothelial cells, but was absent in the tested mouse and human tumor cells. GAPDH was used as an internal control. HUVECs were from Vectec (VT) or Promocell (PC), HPAEC primary human pulmonary aortic endothelial cells, HMVEC primary human microvascular endothelial cells, DU145 human prostate cancer cells, HeLa human cervical cancer

cells, MS1 mouse pancreatic endothelial cell line, 4T1 mouse breast tumor cell line, CT26 mouse colon carcinoma cell line. The production of cAMP in response to extracellular pH changes was assessed c in HUVECS and d in HeLa cells stably transfected with GPR4. Forskolin (FSK, an adenylyl cyclase agonist) was used to increase the assay window in HUVEC cells, see inset. e GPR4specific, but not control siRNAs inhibit the pH-dependent cAMP production (n = 2–3 measurements per point). For all cAMP measurements IBMX (a phophodiesterase inhibitor) was used to stabilize cAMP

growth factor implant angiogenesis model [24–26]. For this purpose, mice were implanted with Teflon chambers containing either VEGF or bFGF, two well known angiogenic factors, or PBS as a baseline control. The

addition of an angiogenic factor triggers the formation of a new, well vascularized tissue around the implanted chamber. As shown in Fig. 2a, b, c, GPR4-deficient mice failed to show an angiogenic response to VEGF, whereas

123

Angiogenesis (2011) 14:533–544

539

Reduced tumor growth in GPR4-deficient mice

the response to bFGF was similar to that observed in wild type controls. In order to confirm this striking effect, we added GPR4-specific siRNAs together with the different growth factors into the chambers and implanted them in wild type mice. This type of local delivery of siRNAs was used before to downregulate angiogenesis targets ([28]; Billy E. et al. manuscript in preparation). In agreement with the previous experiment, two independent GPR4specific siRNAs abrogated the angiogenic effect of VEGF, but had no effect if combined with bFGF (Fig. 2d, e). A similar effect was observed with an siRNA specific for VEGFR2, whereas a control siRNA had no effect with either growth factor.

400 350

GPR4 KO

300

WT

# #

#

250 200

*

150

b

100

#

100 90

ng Tie2 /chamber

tissue weight (mg)

a

Since angiogenesis is crucial for tumors, we next investigated whether lack of GPR4 in the host would affect tumor growth. Syngeneic 4T1 breast tumor cells were implanted into the fat pad of GPR4-deficient or wild type female mice. Tumor growth was monitored over 3 weeks by caliper measurement and tumors were weighed at the end of the experiment. As shown in Fig. 3a tumors were markedly smaller in GPR4-deficient mice as compared to wild type controls. Reduction in tumor size was as strong in the GPR4 KO mice as in wild-type mice treated with the VEGFR2 tyrosine kinase inhibitor PTK787. Also, there

50

80

#

GPR4 KO

#

WT

70

*

60 50 40 30 20 10

0

0 PBS

VEGF

bFGF

PBS

VEGF

bFGF

c

PBS chamber in WT mouse

Empty chamber

tissue weight (mg)

400 350 300 250

#

PBS No siRNA siRNA Ctrl siRNA GPR4 -1

# #

*

150

#

500 450

siRNA GPR4 -2 siRNA VEGFR2

200

e ng Tie2/chamber

d 450

VEGF chamber in GPR4-KO mouse

VEGF chamber in WT mouse

100

400 350

#

300

#

*

250 200 150 100

50

50

0 PBS

VEGF

bFGF

Fig. 2 GPR4 deficiency results in impaired response to VEGF-driven angiogenesis. a Teflon chambers containing agar with or without growth factor were implanted on the back of wild type or GPR4deficient female mice. After 4 days the implant was removed and the tissue which had formed around the chamber was weighed. b The endothelial cell specific marker Tie2 was measured by ELISA as a way to quantify vascularity in the tissue formed around the chamber. c Physical appearance of the different implants. d Teflon chambers

0

PBS

VEGF

bFGF

containing agar with or without growth factor together with siRNAs were implanted on the back of wild type female mice. After 3 days the implant was removed and the tissue which had formed around the chamber was weighed. e The amount of Tie2 marker was measured by ELISA in the tissue formed around the chambers containing growth factor and siRNAs. a, b *P \ 0.001, GPR4 KO versus WT, #P B 0.001 compared to PBS (n = 6 per group); d, e *P \ 0.05 compared to siCtrl, #P B 0.05 compared to PBS (n = 5 per group)

123

540

Angiogenesis (2011) 14:533–544

Tumor volume (mm3)

a

Reduced tumor cell proliferation and changes in vascularity in tumors grown in GPR4-deficient mice

4T1 Tumor

1000

WT PEG

800

WT PTK GPR4 KO PEG

600

GPR4 KO PTK

400 200

#

#

18

22

#

0 8

11

15

Days CT26 Tumor

b

WT

Tumor weight (mg)

c

GPR4 KO

300

CT26 Tumor

250 200 150

#

100 50 0 WT

GPR4 KO

Fig. 3 GPR4 KO mice show reduced tumor growth. a Syngeneic 4T1 breast tumor cells were implanted orthotopically into the fat pad of wild type and GPR4 KO mice. Starting on day 7 mice were treated daily with the VEGFR2 tyrosine kinase inhibitor PTK787 (PTK) at 100 mg/kg or vehicle alone. Tumor growth was measured over 22 days with calipers (n = 6 mice per group, #P \ 0.01 WT vs. GPR4 KO). b Syngeneic CT26 colon tumor cells were implanted orthotopically into the caecum of WT and GPR4 KO mice and animals were sacrificed after 20 days. Example of tumors from WT and GPR4 KO mice. c Weight of CT26 tumors at the end of the experiment (n = 8–10 per group, #P = 0.004 WT vs. GPR4 KO)

was no additive effect on tumor inhibition when GPR4 KO mice were treated with PTK787. Injection of syngeneic CT26 colon tumor cells was used as a second tumor model. Cells were implanted orthotopically into the caecum of mice, and tumor weight was measured after 20 days. In this model the reduction in tumor growth in GPR4-deficient mice compared to wild type controls was even more pronounced (Fig. 3b, c).

123

When CT26 tumor sections were stained for the nuclear proliferation antigen Ki67 a highly significant reduction of proliferating cells in tumors grown in GPR4-deficient mice was observed (Fig. 4a) compared to wild type controls. In contrast, we saw no difference in staining for apoptosis. Representative pictures of cleaved caspase 3 staining are shown in Supplementary Figure 4. It has to be noted that these tumors showed only little apoptosis. After staining of sections for the pan-endothelial marker CD31, we noted that the endothelial cells looked frail and disrupted and vessels appeared not correctly shaped (Fig. 4b). Quantification of vessel density revealed a reduction of this parameter in the tumors grown in the GPR4-deficient mice compared to control, but this difference did not reach statistical significance (WT: 59.4 ± 5.7; GPR4 KO: 44.3 ± 5.3 vessels/mm2 (mean ± sem); P = 0.11). Tomato lectin perfusion also did not show a significant difference between WT and GPR4 KO tumors (WT: 35.78 ± 2.1; GPR4 KO: 27.18 ± 4.2 vessels/mm2 (mean ± sem); P = 0.16). However, when the length of CD31-stained vascular structures was measured, we found that vessels were significantly smaller in tumors grown in GPR4-deficient mice as compared to wild type mice (WT: 48.0 ± 2.3; GPR4 KO: 39.3 ± 1.5 (mean ± sem); P = 0.02), reflecting the disrupted appearance described above. Of note, continuous structures larger than 100 lm in length were more than twice as frequent in tumors taken from wild type compared to GPR4-deficient animals (Fig. 4c, d). Staining of sections for smooth muscle cells/pericytes (desmin, smooth muscle actin and NG2) did not reveal differences between tumors grown in wild type and in GPR4-deficient mice. Representative pictures are shown in Supplementary Figure 5.

GPR4 deficiency results in decreased VEGFR2 levels Trying to understand why GPR4-deficient mice are specifically refractory to VEGF-driven angiogenesis, we investigated the expression of VEGFR2, the main signaling receptor for VEGF on endothelial cells, in wild type and GPR4-deficient mice. Interestingly, we found decreased levels of VEGFR2 in lungs of GPR4-deficient mice (Fig. 5a). This difference was not due to a difference in endothelial cell numbers, since the levels of other endothelial markers such as Tie2, VE-cadherin and EphB4 were similar between wild type and GPR4-deficient mice (Fig. 5b, c). Furthermore, when GPR4-specific siRNAs were transfected into HUVECs, a reduction in VEGFR2

Angiogenesis (2011) 14:533–544

541

surface levels was measured by FACS analysis, which was not seen in HUVECs transfected with a control siRNA (Fig. 5d). We therefore conclude that the reduced angiogenic response to VEGF of GPR4-deficient endothelial cells is due, at least in part, to a decreased expression of VEGFR2.

Discussion We identified GPR4 in an mRNA expression profiling screen as an endothelial cell-specific gene, in accordance with previous reports [5–10]. We further show that in primary human endothelial cells, GPR4 acts as a functional proton-sensing receptor, since the pH-dependent cAMP production could be blocked by GPR4-specific siRNAs and receptor antagonists (Fig. 1 and data not shown). This confirms our earlier observations based on ectopic expression of GPR4 in HEK293 (Human Embryonic Kidney) cells [1]. To further investigate the role of this receptor, GPR4deficient mice were generated. In agreement with the data reported recently by Yang et al. [10] adult mice are viable and fertile and show no major abnormalities in the vascular

b

a Ki67 positive cells per tumor area (%)

30 25 20 15 10

WT

#

5 0 WT

GPR4 KO

GPR4 KO

d

c

45 40

50 40 30 20 10 0 WT

GPR4 KO

Number of vessels

Vessel length (um)

60

*

Fig. 4 Analysis of CT26 tumor sections. a Percentage of proliferating, Ki67-positive, cells in tumors grown in wild type versus GPR4 KO mice (n = 5 tumors per group; #P \ 0.001) b Examples of the fragile and disrupted appearance of blood vessels in CT26 tumors grown in GPR4 KO as compared to wild type mice (CD31 staining, bar = 50 lm) c Average length of CD31 stained vascular structures of tumors grown in WT or GPR4 KO mice (n = 4 per group; *P = 0.022) d Size distribution of CD31-stained vascular structures. Vessel length was determined and values were grouped into 10 lm bins. The number of vessels per lengths bin was plotted (n = 420 for WT, n = 388 for GPR4 KO)

system, indicating that GPR4 does not play a pivotal role in vessel maintenance in an unchallenged situation. In contrast to Yang and collaborators we did not observe indications of reduced litter sizes or a partially penetrant phenotype of vascular malformation in GPR4 KO mice (pure BALB/c or fully backcrossed on B6 background). The reason for these different findings is not known, however, differences in the animal strains used, backcrossing and/or breeding conditions may be relevant. In support of our hypothesis that GPR4 might play a role in angiogenesis, we found that the response to VEGFdriven pathological angiogenesis is markedly reduced in GPR4 KO mice in a growth factor implant model. Furthermore, the addition of GPR4 specific siRNA into the chambers implanted into wild type mice completely abrogated VEGF-induced angiogenesis, whereas bFGF-driven angiogenesis was again unchanged. This result is in agreement with the data obtained in the GPR4-deficent mice and supports the notion that GPR4 is required for VEGF mediated pathological angiogenesis. Interestingly, the response to bFGF is not affected, highlighting a specific interplay between VEGF and acidosis signals. Furthermore, GPR4-deficient mice show reduced tumor growth in two different orthotopic tumor models.

35

GPR4 -KO

30

WT

25 20 15 10 5 0 0

50

100

150

200

250

300

350

Vessel length (um)

123

542

Angiogenesis (2011) 14:533–544

VEGFR2 (ng/mg protein)

a

8 7 6 5

*

4 3 2 1 0

b

800

Tie2 (ng/mg protein)

WT

700

GPR4 KO

600 500 400 300 200 100 0 WT

c

WT

GPR4 KO

GPR4 KO EphB4 VE-Cad Tubulin

d

Fig. 5 GPR4 deficiency results in downregulation of VEGFR2 but not of other endothelial cell markers. VEGFR2 (a) and Tie2 (b) expression was measured in lungs of WT or GPR4 KO mice by ELISA (n = 5 per group, *P = 0.001). c EphB4 and VE-Cadherin (VE-Cad) expression in lung tissue was measured by Western blot (n = 5 per group). Tubulin was used as control for equal protein loading. d HUVEC cells were transfected with different siRNAs and surface expression of VEGFR2 was assessed by FACS analysis 48 h after transfection

The difference in tumor size between WT and GPR4deficient mice was more pronounced in the CT26 tumor model which could be explained by the fact that the CT26 cells secrete about three times more VEGF than the 4T1 cells (143 pg/ml vs. 50 pg/ml medium).

123

The inhibition in tumor growth observed in the 4T1 tumor model was only about 50%, however, inhibition by the VEGFR2 tyrosine kinase inhibitor PTK787/ZK222584 [25] was of similar magnitude in this tumor model. Furthermore, effects of GPR4 deficiency and PTK787/ZK222584 were not additive (Fig. 3a). These data are in line with the concept that compared to CT26, 4T1 tumors are not highly VEGFdependent, and suggest that GPR4 is indeed specifically involved in VEGF-mediated angiogenesis. When tumors were analyzed in more detail, we noted a marked difference in the morphology of the blood vessels between tumors grown in GPR4-deficient versus wild type mice. The vessel staining in the GPR4-deficient tumors was often fragmented, endothelial cells appeared not well aligned and looked fragile (punctuated CD31 staining in Fig. 4b). Indeed, quantification of vessel length revealed that the vascular structures are in average shorter in tumors grown in GPR4-deficient mice when compared to wild type animals. In spite of this, overall vessel density was not dramatically reduced in tumors grown in GPR4-deficient mice. One reason is that due to the morphological appearance of the vessels it was difficult to discriminate between small vessels and fragmented staining of a bigger vessel. This and the high variability between tumors resulted in the observed, but not statistically significant difference in vessel density between tumors grown in GPR4-deficient and wild type mice. Perfusion of mice with tomato lectin did not reveal differences between blood vessels from wild type and GPR4-deficient mice. Since the functionality of blood vessels cannot be fully determined with this technique, it can not be ruled out that the endothelial lining may indeed be more fragile in the GPR4-deficient mice, as suggested also by the phenotype observed by the Witte lab in GPR4deficient mice on mixed background [10]. Finally, a striking decrease in staining for the nuclear proliferation antigen Ki67 was detected in tumors grown in GPR4-deficient compared to wild type animals. This is in agreement with the notion that functional vessels are needed for tumor proliferation as well as with previous reports showing that inhibition of VEGF signaling leads to a decrease in tumor cell proliferation [29, 30]. GPR4-deficient mice have a reduced angiogenic response to VEGF but not to bFGF. This could be explained at least in part by the reduction of VEGFR2 levels which we observed both in the GPR4-deficient mice as well as in endothelial cells treated with GPR4-specific siRNAs. However, since the reduction of VEGFR2 levels is only moderate, we speculate that there may be additional mechanisms accounting for the effect seen in vivo. Particularly, an impact of GPR4 deficiency on endothelial cell signal transduction downstream of VEGFR2 has to be considered. Alternatively, other cell types may play a role. There is evidence that GPR4 is also expressed in some

Angiogenesis (2011) 14:533–544

inflammatory cells and studies in elucidating its function in these cells are ongoing. Inflammatory cells as well as bone marrow-derived cells have been shown to play an important role in angiogenesis. Grunewald et al. [31] reported that bone marrow-derived circulating cells are required for VEGF-driven angiogenesis in several models of adult pathological angiogenesis. Furthermore, it has recently been recognized that infiltrating neutrophils, macrophages and bone marrow-derived myeloid cells can play a crucial role in activating angiogenesis during early stages of carcinogenesis [32–35]. We know that the influx of several types of inflammatory cells is involved in the growth factor implant model (LW unpublished observation), and it can not be excluded that impaired function and/or recruitment of inflammatory cells could contribute to the observed lack of angiogenic response. We hypothesize that in the absence of GPR4, endothelial cells can not fully adapt to the pH change, leading to impaired angiogenesis and tumor growth. Similarly, inflammatory cells, or other stromal cells could be affected. In conclusion, our data indicate that the pH-sensing receptor GPR4 is required for VEGF-driven pathological angiogenesis. While hypoxia induces angiogenesis by increasing VEGF production in tumor cells, extracellular acidosis, sensed by GPR4, may positively modulate the response of endothelial cells to VEGF. In contrast to VEGF- or VEGFR-deficient mice, GPR4-deficient mice are viable and fertile without abnormalities in the vascular system. This indicates that GPR4 is neither required for developmental angiogenesis nor for maintenance of vessels during adulthood. We therefore propose GPR4 as an attractive new therapeutic target for the development of anti-cancer therapies. Acknowledgments We would like to thank Melanie Muller, Juliane Vauxlaire, Corinne Manlius, Marina Maurer, Agnes Feige, Barbara Wilmering-Wetter, Marianne Lemaister, Thierry Doll, Imke RenzAlbrecht and Caterina Safina for excellent technical help, John Monahan for help with Affymetrix analysis, Julie Boisclair for help in characterizing the GPR4 deficient mice, Andreas Theuer for writing the program for Ki67 quantification and Jeanette Wood, Georg Martiny-Baron and Francesco Hofmann for support and critical discussions. Conflict of interest Thomas Suply, Be´range`re Ricoux, Eric Billy, Christian Schnell, Birgit U Baumgarten, Sauveur Michel Maira, Claudia Koelbing, Mireille Ferretti, Bernd Kinzel, Matthias Mu¨ller, Klaus Seuwen and Marie-Gabrielle Ludwig are employees of Novartis AG, Switzerland. Lorenza Wyder is a former employee of Novartis AG.

References 1. Ludwig MG, Vanek M, Guerini D, Gasser JA, Jones CE, Junker U, Hofstetter H, Wolf RM, Seuwen K (2003) Proton-sensing G-protein-coupled receptors. Nature 425:93–98

543 2. Wang JQ, Kon J, Mogi C, Tobo M, Damirin A, Sato K, Komachi M, Malchinkhuu E, Murata N, Kimura T, Kuwabara A, Wakamatsu K, Koizumi H, Uede T, Tsujimoto G, Kurose H, Sato T, Harada A, Misawa N, Tomura H, Okajima F (2004) TDAG8 is a proton-sensing and psychosine-sensitive G-protein-coupled receptor. J Biol Chem 279:45626–456633 3. Ishii S, Kihara Y, Shimizu T (2005) Identification of T cell deathassociated gene 8 (TDAG8) as a novel acid sensing G-proteincoupled receptor. J Biol Chem 280:9083–9087 4. Seuwen K, Ludwig MG, Wolf RM (2006) Receptors for protons or lipid messengers or both? J Recept Signal Transduct Res 26:599–610 5. Lum H, Qiao J, Walter RJ, Huang F, Subbaiah PV, Kim KS, Holian O (2003) Inflammatory stress increases receptor for lysophosphatidylcholine in human microvascular endothelial cells. Am J Physiol Heart Circ Physiol 285:1786–1789 6. Kim KS, Ren J, Jiang Y, Ebrahem Q, Tipps R, Cristina K, Xiao YJ, Qiao J, Taylor KL, Lum H, Anand-Apte B, Xu Y (2005) GPR4 plays a critical role in endothelial cell function and mediates the effects of sphingosylphosphorylcholine. FASEB J 19:819–821 7. Qiao J, Huang F, Naikawadi RP, Kim KS, Said T, Lum H (2006) Lysophosphatidylcholine impairs endothelial barrier function through the G protein-coupled receptor GPR4. Am J Physiol Lung Cell Mol Physiol 291:91–101 8. Huang F, Mehta D, Predescu S, Kim KS, Lum H (2007) A novel lysophospholipid- and pH-sensitive receptor, GPR4, in brain endothelial cells regulates monocyte transmigration. Endothelium 14:25–34 9. Zou Y, Kim CH, Chung JH, Kim JY, Chung SW, Kim MK, Im DS, Lee J, Yu BP, Chung HY (2007) Upregulation of endothelial adhesion molecules by lysophosphatidylcholine. Involvement of G protein-coupled receptor GPR4. FEBS J 274:2573–2584 10. Yang LV, Radu CG, Roy M, Lee S, McLaughlin J, Teitell MA, Iruela-Arispe ML, Witte ON (2007) Vascular abnormalities in mice deficient for the G protein-coupled receptor GPR4 that functions as a pH sensor. Mol Cell Biol 27:1334–1347 11. Bergers G, Benjamin LE (2003) Tumorigenesis and the angiogenic switch. Nat Rev Cancer 3:401–410 12. Heath VL, Bicknell R (2009) Anticancer strategies involving the vasculature. Nat Rev Clin Oncol 6:395–404 13. Griffiths JR (1991) Are cancer cells acidic ? Br J Cancer 64:425–427 14. Gatenby RA, Gillies RJ (2004) Why do cancers have high aerobic glycolysis? Nat Rev Cancer 4:891–899 15. Chiche J, Brahimi-Horn MC, Pouyssegur J (2010) Tumor hypoxia induces a metabolic shift causing acidosis: a common feature in cancer. J Cell Mol Med 14:771–794 16. Gambhir SS (2002) Molecular imaging of cancer with positron emission tomography. Nat Rev Cancer 2:683–693 17. Vander Heiden MG, Cantley LC, Thompson CB (2009) Understanding the Warburg effect: the metabolic requirements of cell proliferation. Science 324:1029–1033 18. Park HJ, Lyons JC, Ohtsubo T, Song CW (1999) Acidic environment causes apoptosis by increasing caspase activity. Br J Cancer 80:1892–1897 19. Rofstad EK, Mathiesen B, Kindem K, Galappathi K (2006) Acidic extracellular pH promotes experimental metastasis of human melanoma cells in athymic nude mice. Cancer Res 66:6699–6707 20. DeClerck K, Eble RC (2010) The role of hypoxia and acidosis in promoting metastasis and resistance to chemotherapy. Front Biosci 15:213–225 21. Brahimi-Horn MC, Pouyssegur J (2007) Hypoxia in cancer cell metabolism and pH regulation. Essays Biochem 43:165–178 22. Salomon Y (1979) Adenylate cyclase assay. Adv Cyclic Nucleotide Res 10:35–55

123

544 23. Hu¨sken D, Asselbergs F, Kinzel B, Natt F, Weiler J, Martin P, Ha¨ner R, Hall J (2003) mRNA fusion constructs serve in a general cell-based assay to profile oligonucleotide activity. Nucleic Acids Res 31:102 24. Wood J, Bonjean K, Ruetz S, Bellahce`ne A, Devy L, Foidart JM, Castronovo V, Green JR (2002) Novel antiangiogenic effects of the bisphosphonate compound zoledronic acid. J Pharmacol Exp Ther 302:1055–1061 25. Wood JM, Bold G, Buchdunger E, Cozens R, Ferrari S, Frei J, Hofmann F, Mestan J, Mett H, O’Reilly T, Persohn E, Rosel J, Schnell C, Stover D, Theuer A, Towbin H, Wenger F, WoodsCook K, Menrad A, Siemeister G, Schirner M, Thierauch KH, Schneider MR, Drevs J, Martiny-Baron G, Totzke F (2000) PTK787/ZK 222584, a novel and potent inhibitor of vascular endothelial growth factor receptor tyrosine kinases, impairs vascular endothelial growth factor-induced responses and tumor growth after oral administration. Cancer Res 60:2178–2189 26. Martiny-Baron G, Holzer P, Billy E, Schnell C, Brueggen J, Ferretti M, Schmiedeberg N, Wood JM, Furet P, Imbach P (2010) The small molecule specific EphB4 kinase inhibitor NVPBHG712 inhibits VEGF driven angiogenesis. Angiogenesis 13:259–267 27. Ehrbar M, Djonov VG, Schnell C, Tschanz SA, Martiny-Baron G, Schenk U, Wood J, Burri PH, Hubbell JA, Zisch AH (2004) Cell-demanded liberation of VEGF121 from fibrin implants induces local and controlled blood vessel growth. Circ Res 94:1124–1132 28. Chae SS, Paik JH, Furneaux H, Hla T (2004) Requirement for sphingosine 1-phosphate receptor-1 in tumor angiogenesis demonstrated by in vivo RNA interference. J Clin Invest 114:1082–1089

123

Angiogenesis (2011) 14:533–544 29. Fiedler W, Serve H, Do¨hner H, Schwittay M, Ottmann OG, O’Farrell AM, Bello CL, Allred R, Manning WC, Cherrington JM, Louie SG, Hong W, Brega NM, Massimini G, Scigalla P, Berdel WE, Hossfeld DK (2005) A phase 1 study of SU11248 in the treatment of patients with refractory or resistant acute myeloid leukemia (AML) or not amenable to conventional therapy for the disease. Blood 105:986–993 30. Wedam SB, Low JA, Yang SX, Chow CK, Choyke P, Danforth D, Hewitt SM, Berman A, Steinberg SM, Liewehr DJ, Plehn J, Doshi A, Thomasson D, McCarthy N, Koeppen H, Sherman M, Zujewski J, Camphausen K, Chen H, Swain SM (2006) Antiangiogenic and antitumor effects of bevacizumab in patients with inflammatory and locally advanced breast cancer. J Clin Oncol 24:769–777 31. Grunewald M, Avraham I, Dor Y, Bachar-Lustig E, Itin A, Jung S, Chimenti S, Landsman L, Abramovitch R, Keshet E (2006) VEGF-induced adult neovascularization: recruitment, retention, and role of accessory cells. Cell 124:175–189 32. Coffelt SB, Lewis CE, Naldini L, Brown JM, Ferrara N, De Palma M (2010) Elusive identities and overlapping phenotypes of proangiogenic myeloid cells in tumors. Am J Pathol 176:1564–1576 33. Tazzyman S, Lewis CE, Murdoch C (2009) Neutrophils: key mediators of tumour angiogenesis. Int J Exp Pathol 90:222–231 34. Murdoch C, Muthana M, Coffelt SB, Lewis CE (2008) The role of myeloid cells in the promotion of tumour angiogenesis. Nat Rev Cancer 8:618–631 35. Coffelt SB, Hughes R, Lewis CE (2009) Tumor-associated macrophages: effectors of angiogenesis and tumor progression. Biochim Biophys Acta 1796:11–18