Abstract
Solid tumor growth is intimately associated with angiogenesis, a process that is efficiently triggered by hypoxia. Therefore, oxygen-sensitive signaling pathways are thought to play a critical role in tumor angiogenesis and progression. Here, the function of prolyl hydroxylase-4 (PHD4), a relative of the prolyl hydroxylase domain proteins 1–3 that promote the degradation of hypoxia-inducible factors (HIF), was interrogated. To test the hypothesis that PHD4 might inhibit tumor angiogenesis, it was overexpressed in osteosarcoma cells, and unexpectedly, this manipulation led to increased tumor blood vessel density. However, the newly formed blood vessels were smaller than normal and appeared to be partially nonfunctional, as indicated by poor vessel perfusion. PHD4 overexpression in tumor cells stimulated the expression of TGF-α, which was necessary and sufficient to promote angiogenic sprouting of endothelial cells. On the other hand, PHD4 overexpression reduced HIF-2α protein levels, which in turn inhibited in vivo tumor growth. Combined, elevated PHD4 levels deregulate angiogenesis via increased TGF-α expression in vitro and in vivo. These data support the hypothesis that tumor growth can be uncoupled from vessel density and that the individual PHD family members exert distinct functions in tumors.
Implications: PHD4 influences tumor growth and vascularization through discrete mechanisms and molecular pathways that likely have therapeutic potential. Mol Cancer Res; 11(11); 1337–48. ©2013 AACR.
Introduction
A characteristic feature of solid cancer progression is low tissue oxygen tension (hypoxia) within the tumor, which results from rapid tumor-cell proliferation and insufficient blood supply. Hypoxic cells activate adaptive responses including increased cell survival, changes in metabolism, and induction of angiogenesis (1, 2). Central to the cellular hypoxia response is the transcriptional regulator hypoxia-inducible factor (HIF)-1. HIF-1 stimulates the expression of genes with important roles in cancer, such as VEGF (2–4). HIF-1 and the other members of the HIF family (HIF-2 and HIF-3) are heterodimeric proteins consisting of an oxygen-sensitive α-subunit and a constitutive β-subunit. Whereas HIF-1α is a ubiquitous protein that accumulates in all hypoxic cells, HIF-2α is expressed in a cell type–specific manner, for example in the developing vasculature, and exerts more specific functions (5–8). Although HIF-1 and HIF-2 regulate an overlapping set of target genes, they have distinct physiologic functions in vivo (9).
HIF-1α and HIF-2α levels are elevated in many human cancers (10, 11), which correlates with enhanced invasive and metastatic potential of tumor cells and poor prognosis (12). Several studies have shown that HIF-1α loss-of-function results in impaired tumor growth and angiogenesis (13, 14). HIF-1 is therefore considered a potential target for anticancer therapy (15). However, HIF-1α deficiency can also enhance the growth of certain tumors by preventing apoptotic cell death (13, 16), and HIF-2α can also function as tumor suppressor gene (17, 18). Thus, although it is clear that HIF-1 and HIF-2 play important roles in tumor pathophysiology, their functions seem to be tumor type–dependent.
HIF activity in tumors is modulated through multiple mechanisms, but a critical regulatory step is the oxygen-dependent control of HIF-α stability. In normoxic cells, the HIF-α subunit is rapidly degraded via the proteasome, a process that is initiated by hydroxylation of conserved proline residues by prolyl 4-hydroxylase domain (PHD) proteins (1). These enzymes function as cellular oxygen sensors, as they require molecular oxygen for their activity. When oxygen levels are low, PHD proteins are less active and HIF proteins are not targeted for degradation. This leads to the accumulation of HIF dimers in the cell nucleus, where they activate the transcription of a large number of target genes involved in physiologic and pathologic processes (1).
The PHD family comprises three bona fide HIF prolyl hydroxylases (PHD1, PHD2, and PHD3) and the prolyl hydroxylase-4 (PHD4; also called PH-4 or P4H-TM; refs. 19–22). PHDs are ubiquitously expressed, but their expression levels in tissues and intracellular localizations vary (23, 24). PHD2 has been proposed to be the key oxygen sensor for setting steady-state levels of HIF-1α (25, 26). The prominent physiologic function of PHD2 is underscored by the observation that knockout of the PHD2 gene leads to embryonic lethality, whereas PHD1- and PHD3-deficient mice are viable (27). Although PHDs are thought to primarily control HIF stability, recent evidence shows that they influence also other signaling pathways, including the NF-κB pathway (28), and that certain functions of PHD2 do not involve hydroxylation (29).
To date, relatively little is known about the ability of the individual members of the PHD family to interfere with tumor growth and angiogenesis (30). PHD1 overexpression in colon carcinoma cells inhibits the vascularization and growth of solid tumors in mice (31). It was reported that PHD2 expression is reduced in human colon carcinoma compared with normal colon tissue and that silencing of PHD2 in human colon carcinoma cell lines enhances neovascularization and growth of tumor xenografts in mice, consistent with a function for PHD2 as negative regulator of tumor angiogenesis and progression (29). In contrast, we observed recently that PHD2 silencing can inhibit growth of various murine tumors despite increasing angiogenesis (32). Thus, PHD2 seems to limit angiogenesis in tumors, but its effect on tumor growth can be tumor type–dependent.
Here, we investigated the effect of PHD4 on tumor growth and vascularization in a well-characterized mouse osteosarcoma model (32–34). A unique feature of PHD4 is that it possesses a transmembrane domain and is localized to the endoplasmic reticulum (21, 22). PHD4 hydroxylates the two critical prolines in the HIF-1α oxygen-dependent degradation domain in vitro and suppresses the HIF transactivation activity (21, 22). PHD4 mRNA is expressed in human tissues and various tumor types (22); however, its function in cancer is poorly studied. Here, we show that overexpression of PHD4 led to decreased HIF-2α protein levels and greatly reduced experimental LM8 osteosarcoma growth in mice. Unexpectedly, however, tumor growth inhibition was accompanied by increased vessel density, which was caused by upregulation of TGF-α. On the other hand, silencing of HIF-2α in LM8 cells led to tumor growth inhibition without influencing tumor vessel density. Taken together, these observations suggest that elevated PHD4 levels disturb the angiogenic balance in osteosarcoma via induction of the TGF-α pathway and inhibit tumor growth by reducing the expression of HIF-2α.
Materials and Methods
Cell culture and generation of stable cell clones
LM8 osteosarcoma cells (35) were obtained from Prof. Christian Beltinger, University Clinic Ulm, Germany, in 2005. Cells were cultured in Minimum Essential Medium (MEM-α; Gibco) supplemented with 1% l-glutamine, 1% nonessential amino acids (NEAA), and 10% fetal calf serum (FCS). Embryonic endothelial progenitor cells (eEPC) were maintained in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 1% l-glutamine, 1% NEAA, 0.0007% β-mercaptoethanol, and 20% FCS. Growth in syngeneic C3H mice and MMP-2 expression confirmed identity of LM8 cells. Culture times were less than 6 months per batch of cells. Human umbilical vein endothelial cells (HUVEC; Lonza) were grown in endothelial cell growth medium (EGM-2 Bulletkit; Clonetics). Incubation in hypoxic atmosphere was performed in a modular incubator chamber (Billups-Rothenberg Inc.) using premixed gases (University Clinic, TU Dresden). To generate stable LM8 and eEPC lines overexpressing human PHD4, cells were transfected with a full length hPHD4 cDNA clone in a pcDNA3.1 vector using transfection reagent (Lipofectamine2000; Invitrogen) and selected in medium supplemented with 0.5 mg/mL G418. Cells transfected with empty pcDNA3.1 vector served as a control for all experiments.
Silencing of TGF-α or HIF-2α
LM8 clones silenced for TGF-α were generated essentially as described in refs. (32, 34). In short, recombinant pLVHTM vector encoding shRNA specific for murine TGF-α was generated and transfected into 293T cells, together with plasmids psPAX2 and pMD2.G (kind donation of Dr. D. Trono, Geneva, Switzerland). The oligonucleotide encoding shRNA specific for TGF-α was: 5′-CCC ACA CTC AGT ACT GCT T. Vector containing a scrambled (scr) sequence, 5′-AGT CGC TTA GAA ACG AGA A, was used as a control. Subsequently, LM8-hPHD4 cells were transduced with recombinant lentiviral particles and stable clones expressing the pLVTHM-encoded GFP were selected. LM8-pcDNA3 clones silenced for HIF-2α were generated by transducing LM8-pcDNA3 cells with lentiviral particles encoding shRNA specific for HIF-2α (5′-CTC AGT TAC AGC CAC TCG TCA CTG).
In vitro angiogenesis assay
The assay used is based on established procedures (36, 37), with modifications described later. In brief, HUVECs were seeded on dextran-coated Cytodex 3 microcarrier beads (Amersham Pharmacia Biotech) and beads were embedded in fibrin gel in 96-well plates. To study paracrine interactions between tumor and endothelial cells, a feeder layer consisting of 5,000 LM8 cells per well, suspended in 200 μL EGM-2 medium, was seeded on top of the fibrin gels. Plates were incubated at 37°C in 5% CO2 and 95% relative humidity. Recombinant TGF-α (R&D Systems) or the EGF receptor (EGFR) blocking antibody (clone 225; Lab Vision) were used at the indicated concentrations. After 3 days, the 96-well plates were washed three times with PBS and fixed in 4% paraformaldehyde overnight. Quantitative analysis was performed using the Analysis imaging software (Olympus) by measuring the number and the length of the sprouts and the number of the branch points. Approximately 50 beads per experimental group were analyzed in at least two independent experiments.
Reverse transcription PCR analysis
Total RNA was isolated using RNeasy Mini Kit and RNAse-free DNase set (Qiagen). Aliquots of 2 μg RNA were reverse transcribed using Superscript II reverse transcriptase (Invitrogen) and random hexameric primers (Roche). Reverse transcription (RT)-PCR was performed with the GoTaq DNA Polymerase (Promega). PCR primers were from MWG Biotech: mActin-forw 5′-TTC CGA TGC CCT GAG GCT CT; mActin-rev 5′-CAG GAG GAG CAA TGA TCT TG; hPHD4-forw 5′-GGC GCA GAT GAA GGG GTT ACA G; hPHD4-rev 5′-CAG GCG ACA GGC GAG TGA GG; mTGF-α-forw 5′-CTC TGG GTA CGT GGG TGT TC, and mTGF-α-rev 5′-GAG CTG ACA GCA GTG GAT CA.
Western blot analysis
To prepare lysates, cells were washed twice with cold PBS and scraped on ice in lysis buffer (1% NP40, 0.5% sodium deoxycholate, and 0.1% SDS) supplemented with complete Protease Inhibitor Cocktail (Roche). Tumor tissue was cut into small pieces and homogenized in lysis buffer by passing through a needle (20G, 15 times) followed by centrifugation at 13,600 ×g for 15 minutes. The supernatant was transferred to a fresh tube and protein concentration was determined by Bradford assay (Pierce). Proteins were separated by SDS-PAGE and transferred onto nitrocellulose (Schleicher and Schuell). The membranes were probed with a primary antibody, followed by incubation with horseradish-peroxidase–conjugated secondary antibody. The antibodies used were: rb-anti-hPHD4, rb-anti-HIF-2α, TGF-α (Novus Biologicals), rb-anti-HIF-1α (Cayman), and rb-anti-β-actin (Sigma). Immunoreactivity was detected using an enhanced-chemiluminescence (ECL) detection system (Pierce).
Tumor experiments
Tumor cells (1 × 106 cells per tumor) were injected subcutaneously into the right and left hind limbs of 8-to 10-week-old female C3H mice (Taconic) as described in refs. (32, 34). Tumor area was measured every other day starting from day 7. The tumor volume was calculated as (a2 × b)/2, where b is the larger diameter of the tumor and a is the largest diameter perpendicular to b. All experiments were approved by and performed according to the guidelines of the animal ethical committee of the Medical Faculty Carl Gustav Carus, TU Dresden, Germany. To analyze vessel perfusion, Hoechst 33342 dye (0.8 mg in 100 μL; Sigma) was injected into the tail vein 1 minute before the tumors were isolated.
Immunohistology
Tumor vessels were visualized on frozen sections by immunohistochemistry using rat-anti-platelet-endothelial cell adhesion molecule (PECAM) antibody (clone Mec13.3; ref. 38) and the Vectastain ABC and Peroxidase Substrate AEC kits according to manufacturer's instructions (Vector Laboratories). Control staining was done by omitting the primary antibody. Sections were counterstained with hematoxilin and eosin (H&E) solution. To analyze vessel perfusion in tumors, Hoechst 33342-injected LM8 tumors were stained by immunofluorescence for PECAM, using goat anti-rat Alexa 594 (Molecular Probes) as secondary antibody. Images were taken using an Axioplan2 microscope and the AxioVision AC software (Carl Zeiss) and digitally processed using ImageJ software. Viable and necrotic tumor areas were defined by morphologic criteria on H&E-stained sections. Endothelial area and vessel density were determined by morphometry on whole tumor sections each, obtained from different regions of at least seven different tumors. From each slide, pictures were taken from the entire viable tumor tissue area at 25× or 100× magnification and the vessels were counted to determine the vessel density (= number of vessels per area). The endothelial area was calculated using the ImageJ software (NIH).
Affymetrix microarray
Expression profiling of tumor cells was performed as described previously (39). Fragmented biotin-labeled cRNA was prepared from total RNA isolated from LM8-hPHD4 and control cells and hybridized to MOE430A GeneChip arrays as recommended by manufacturer (Affymetrix). Arrays were scanned with a GeneChip Scanner 2500 and the images were quantified using MAS 5.0 software (MicroArray Suite; Affymetrix). Datasets were submitted to Gene Expression Omnibus (GEO) at National Center for Biotechnology Information (NCBI; entry GSE49680). The standard guidelines for Minimum Information About a Microarray Gene Experiment (MIAME guidelines) were followed.
Statistical analysis
All statistical analyses were performed using SigmaPlot and GraphPad Prism 4.02 Software. P values less than 0.05 were considered statistically significant. Tumor-associated variables were tested for statistical significance with a two-tailed Student t test.
Real-time PCR
RNA was purified from cell lysates using the Universal RNA Purification Kit (Roboklon GmbH), or from tumor tissue lysates using an RNeasy Mini Kit (Qiagen). Aliquots of 1 to 2 μg of total RNA were reverse transcribed using SuperScript II reverse transcriptase (Life Technologies GmbH). Of note, 10 ng cDNA per reaction was subjected to quantitative PCR using the Maxima SYBR Green qPCR Master Mix (K0252; Fermentas GmbH). The experiments were performed on an iCycler iQ (Bio-Rad Laboratories GmbH). TATA box-binding protein (TBP) as well as eucaryotic elongation factor 2 (eEF2) genes were used as reference. Oligonucleotide primers were from MWG Biotech: mTGF-α-forw, 5′-CTC TGG GTA CGT GGG TGT TC; mTGF-α-rev, 5′-GAG CTG ACA GCA GTG GAT CA; eEF2-forw, 5′-ATC CTC ACC GAC ATC ACC AAG; eEF2-rev, 5′-CTG CTC TGG ACA CTG GAT CTC; mTBP-forw, 5′-TCT ACC GTG AAT CTT GGC TGT AAA; mTBP-rev, 5′-TTC TCA TGA TGA CTG CAG CAA A.
Results
Generation and characterization of cell clones overexpressing human PHD4
The aim of this study was to investigate the effect of PHD4 overexpression on tumor angiogenesis and progression. First, we stably transfected LM8 osteosarcoma cells with a human PHD4 (hPHD4) cDNA and analyzed the resulting cell clones for PHD4 expression by Western blot and RT-PCR analysis. Two clones that expressed high levels of PHD4 (LM8-hPHD4#11 and #14; Fig. 1A) were used for subsequent experiments. These cell clones did not differ in their in vitro proliferation rate from control clones transfected with empty vector (cell number doubling time in hours: control, 20.5 ± 4.0; LM8-hPHD4#11, 17.9 ± 2.1; and LM8-hPHD4#14, 21.5 ± 5.3). PHD4 overexpression did not cause obvious morphologic changes of the cells (data not shown). We also generated transfectants of murine eEPCs to study the influence of PHD4 overexpression on HIF stability in a second cell type.
Overexpression of PHD4 in LM8 osteosarcoma cells reduces HIF-2α levels. A (left), expression of human PHD4 protein was analyzed by Western blot in lysates from LM8 cells stably transfected with hPHD4 cDNA (LM8-hPHD4 clones #11 and #14) and control LM8 cells (LM8-pcDNA3). β-actin was used as loading control. The right panel shows the quantitation of hPHD4 mRNA levels by quantitative reverse transcription PCR (qRT-PCR). B (left), Western blot detection of HIF-1α and β-actin in cell extracts from control LM8 cells and LM8 cells overexpressing PHD4 (clone #14) under normoxic or hypoxic conditions (24 hours, 1% O2). Right, HIF-2α protein was detected in hypoxic LM8 cells only. LM8 cells overexpressing PHD4 (clone #11) showed reduced HIF-2α levels as compared with control LM8 cells.
To examine the effect of PHD4 overexpression on HIF stability, LM8-hPHD4 cells and control LM8 cells were cultured under normoxic conditions and cell lysates were analyzed by Western blot for HIF-1α and HIF-2α proteins. Expression of HIF-1α was induced in LM8 cells following exposure to 1% O2 for 24 hours (Fig. 1B). PHD4 overexpression only slightly diminished HIF-1α protein levels. HIF-2α protein was detected in hypoxic LM8 control cells or LM8-hPHD4 cells (Fig. 1B) but not in normoxic cells (not shown). Hypoxic LM8-hPHD4 cells showed reduced HIF-2α levels as compared with hypoxic LM8 control cells. Downregulation of HIF-2α by PHD4 overexpression was also observed in murine eEPCs that express abundant HIF-2α levels (Supplementary Fig. S1).
Overexpression of PHD4 in LM8 osteosarcoma inhibits subcutaneous tumor growth
We next analyzed whether PHD4 overexpression influences tumor growth in vivo. LM8-hPHD4 cells and control LM8 cells were injected subcutaneously and solid tumor growth was monitored over a 3-week period. We observed a strong and highly significant reduction in tumor growth in animals injected with LM8-hPHD4 cells compared with controls (Fig. 2A and B).
PHD4 overexpression in LM8 osteosarcoma inhibits tumor growth. A and B, control LM8 cells and LM8 cells overexpressing PHD4 (clones #11 and #14) were injected subcutaneously into C3H mice (n = 6–7 per group). Tumor volume was determined every 2 to 3 days. The data represent the mean tumor volume per group. The results of one representative experiment out of three are shown. Asterisks indicate a significant (*, P < 0.05; ***, P < 0.001) difference between control tumors and tumors overexpressing PHD4 (clones #11 and #14). C, necrotic area was determined on H&E-stained sections. No significant difference between control and PHD4-overexpressing LM8 tumors was observed. D, the proportion of proliferating tumor tissue was evaluated after staining for the proliferation antigen Ki-67 staining. Each bar represents the mean ± SEM of 5 to 6 analyzed tumors (**, P < 0.01).
LM8 tumors overexpressing PHD4 exhibit more, but smaller, vessels with partially impaired functionality
It was previously reported that the overexpression of PHD1 in colon carcinoma cells results in reduced tumor growth, which was caused by increased necrosis and inhibition of angiogenesis (31). To analyze whether PHD4 overexpression would have similar consequences, cryosections of LM8-hPHD4 tumors were prepared and stained with H&E. Histologic analysis of the tumors revealed extensive necrotic areas, but no significant difference in the extent of necrosis between LM8-hPHD4 and LM8-pcDNA3 control tumors was observed (Fig. 2C). Consistent with the idea that cell death is not increased in LM8-hPHD4 tumors, caspase-3 staining of frozen tumor sections did not reveal obvious differences in the number of apoptotic cells between LM8-hPHD4 tumors and LM8-pcDNA3 control tumors (data not shown). In contrast, cell proliferation, as determined by Ki-67 staining, was significantly decreased in hPHD4-overexpressing tumors (Fig. 2D). Thus, the reduced growth of LM8-hPHD4 tumors is not caused by necrotic cell death, but rather by reduced tumor cell proliferation in vivo.
It is well established that hypoxia stimulates the expression of VEGF and other angiogenic factors via the HIF pathway in tumor cells leading to the induction of tumor angiogenesis (2, 3). We therefore initially hypothesized that overexpression of PHD4 might inhibit the production of angiogenic factors and impede tumor angiogenesis. To test this hypothesis, we determined the endothelial area and vessel density in sections of LM8 tumors after immunohistochemical staining for the endothelial marker PECAM (Fig. 3A). Analysis of the relative endothelial area in the viable tumor tissue revealed no significant difference between LM8-PHD4 tumors and control tumors (Supplementary Fig. S2). However, closer examination showed that LM8 osteosarcomas overexpressing PHD4 contained more, but considerably smaller vessels, than control tumors (Fig. 3A and B).
PHD4 overexpression in LM8 osteosarcoma stimulates nonproductive angiogenesis. A, immunohistochemical staining for PECAM on sections of control or PHD4-overexpressing LM8 tumors (clone #11; magnification ×100). B, the total number of vessels per area was counted. Each bar represents the mean ± SEM of 7 analyzed tumors. Asterisks indicate a significant difference (***, P < 0.001) between control tumors and tumors overexpressing PHD4 (clone #11). C, double immunofluorescence for PECAM (red) and Hoechst 33342 (blue) on sections of control or PHD4-overexpressing LM8 tumors (clone #11). D, quantification of nonperfused tumor vessels. Each bar represents the mean ± SEM of 5 analyzed tumors (**, P < 0.01). All results were confirmed with the second PHD4-overexpressing cell clone (#14).
One possible explanation for this seemingly paradoxical result, that tumor growth is inhibited while vascularity is increased, would be that the functionality of blood vessels in LM8-hPHD4 tumors is impaired. To address this question, we analyzed perfusion of tumor tissue by monitoring the distribution of a fluorescent Hoechst dye after intravenous injection. Tumor sections were prepared and stained by immunofluorescence for PECAM to visualize the tumor endothelium. Well-perfused tumor areas show blue staining with Hoechst 33342, indicating the presence of functional tumor vessels. In LM8-hPHD4 osteosarcomas, large tumor areas were not perfused, although they contained many PECAM-positive vessels (Fig. 3C). This was not the case in control LM8 tumors, in which areas with reduced perfusion were typically avascular. As shown in Fig. 3D, the proportion of nonperfused vessels was significantly increased in LM8-hPHD4 tumors. These results show that although PHD4 overexpression in LM8 osteosarcoma stimulates angiogenesis, a significant proportion of the newly formed vessels is not functional.
Overexpression of PHD4 in LM8 cells leads to the upregulation of the proangiogenic factor TGF-α
We next asked which factors produced by LM8-hPHD4 cells are responsible for the increased, partially nonproductive angiogenesis. One possible candidate is VEGF, a key regulator of tumor angiogenesis, which is upregulated by HIF-1 (3). However, we detected no significant increase in VEGF expression in LM8-hPHD4 cells or in lysates from subcutaneous tumors by ELISA and RT-PCR (Supplementary Fig. S3). These data indicate that the increased vessel density in the tumors overexpressing PHD4 is not caused by VEGF. A systematic search for angiogenic factors that are upregulated in LM8-hPHD4 cells was performed by Affymetrix Microarray Expression profiling. Consistent with the results of the ELISA and RT-PCR analyses, no change in VEGF mRNA expression was observed in LM8-hPHD4 cells compared with control cells. Yet, in addition to various genes that have not been implicated in angiogenesis, the gene encoding the angiogenic factor TGF-α (40) was robustly upregulated in both LM8-hPHD4 clones (fold change, 3.25; P < 0.003). This result was confirmed by RT-PCR (Fig. 4A) on cells and on tumor lysates (Fig. 4B). These data suggest that TGF-α might promote angiogenesis in PHD4-overexpressing LM8 tumors.
Overexpression of PHD4 in LM8 osteosarcomas results in increased TGF-α expression. A, mRNA expression levels of human PHD4 and TGF-α in cell extracts from control and PHD4-overexpressing LM8 cells (#11 and #14) cultured under normoxic and hypoxic condition (6 hours, 1% O2). β-actin was used as loading control. B, RT-PCR detection of mRNA for human PHD4, TGF-α, and β-actin in tumor lysates from control LM8 and LM8 cells overexpressing PHD4 (#11). C, quantification of TGF-α mRNA levels in tumor lysates by qRT-PCR. Each bar represents the mean ± SEM of 5 analyzed tumors (***, P < 0.001). D, Western blot analysis of lysates from LM8-hPHD4 cells and LM8 control cells detects a ≈30 kD TGF-α precursor. Specificity of staining was confirmed by reduced signal intensity in lysates of LM8-hPHD4 cells silenced for TGF-α (Supplementary Fig. S4).
Overexpression of PHD4 in LM8 cells stimulates vessel branching in vitro
Next, we performed an in vitro angiogenesis assay that allowed us to monitor the production of angiogenic factors by PHD4-overexpressing LM8 tumor cells. In this assay, HUVECs grown on microbeads were embedded in fibrin gels and, subsequently, LM8 osteosarcoma cells (with or without hPHD4 expression vectors) were seeded on top of the gels. In this setting, diffusible factors secreted by the tumor cells stimulate endothelial cell growth and sprouting in a paracrine fashion. After 1 to 3 days, individual beads were analyzed for sprout number, sprout length, and the number of branch points giving rise to side sprouts. No significant difference in the number or length of endothelial sprouts was observed between LM8-hPHD4 cells and control cells. In contrast, the number of branch points was significantly increased in assays with LM8-hPHD4 cells (Fig. 5A and B). This shows that overexpression of PHD4 in LM8 osteosarcoma cells leads to the production of factors that enhance endothelial cell branching, consistent with our observation that LM8-hPHD4 tumors harbor significantly more and smaller vessels than controls.
Overexpression of PHD4 in LM8 increases endothelial branching via TGF-α/EGFR in an in vitro angiogenesis assay. A, representative images of HUVEC sprouts grown in a fibrin gel and exposed to control or PHD4-overexpressing LM8 cells (#11 and #14; magnification ×100). B, quantification of the total number of endothelial branching points. Each bar represents the mean ± SEM of approximately 20 analyzed beads. Asterisks indicate a significant (***, P < 0.001) difference between HUVECs exposed to control and hPHD4-overexpressing LM8 cells (#11 or #14). B, quantification of the total number of branch points in sprouts formed by HUVECs in fibrin gels. C, effect of recombinant TGF-α on endothelial branching. LM8 transfected with control vector were grown on top of the gels in the presence or absence of recombinant TGF-α (0.25, 0.5, or 1 ng/well). LM8-hPHD4 (#11) cells were used as a control and stimulated branching to a similar extent as TGF-α. Each bar represents the mean ± SEM of approximately 20 analyzed beads. Asterisks indicate a significant (***, P < 0.001) difference in vessel branching between control and TGF-α–treated cultures. D, effect of EGFR inhibition on endothelial branching induced by PHD4-overexpressing tumor cells. Quantification of the total number of endothelial branch points observed after exposure of HUVEC cultures on microbeads to control LM8 cells, or hPHD4-overexpressing LM8 cells (hPHD4#11) upon addition of control IgG1 antibody or 225 EGFR blocking antibody (at concentrations of 0.1, 0.2, or 1 μg/mL). Asterisks indicate a significant (***, P < 0.001) difference in vessel branching between cultures treated with the control antibody and with the EGFR-blocking antibody. E, effect of TGF-α knockdown in PHD4-overexpressing tumor cells on endothelial branching. Quantification of the total number of endothelial branch points observed after exposure of HUVEC cultures on microbeads to control LM8-hPHD4 cells (hPHD4#11-shscr9) and LM8-hPHD4 cells silenced for TGF-α. Asterisks indicate a significant (**, P < 0.01; ***, P < 0.001) difference in vessel branching.
Endothelial branching is stimulated by TGF-α/EGFR signaling
To address the question whether the observed increase in endothelial branching is caused by TGF-α, we performed an in vitro angiogenesis assay in which recombinant TGF-α (at concentrations of 1, 0.5, or 0.25 ng/well) was added to a feeder layer of control LM8 cells. As with PHD4-overexpressing LM8 cells, the application of TGF-α did not alter the number or the length of endothelial sprouts, but increased the number of branch points at concentrations above 0.5 ng per well (Fig. 5C). These results suggest that TGF-α is the main factor produced by LM8-hPHD4 cells that stimulates endothelial branching. To test this hypothesis, we inhibited TGF-α activity in the angiogenesis assay. Because TGF-α signals via the EGFR on endothelial cells (41), we blocked its activity using an EGFR-specific antibody. This treatment reduced the number of branch points in a concentration-dependent manner, down to the level of control LM8 cells (Fig. 5D). Staining of tumor sections with antibodies specific for phosphorylated EGFR revealed that the amount of pEGFR was increased in LM8-hPHD4 tumors as compared with control LM8 tumors (not shown). The staining pattern was consistent with increased activation of this receptor in tumor cells and the tumor vasculature.
We also silenced TGF-α expression in LM8-hPHD4 cells by lentiviral transduction with shRNA against TGF-α (see Supplementary Fig. S4). Three independent LM8-shTGFα clones and a control clone were analyzed in sprouting assays and showed significantly reduced angiogenic activity (Fig. 5E). Taken together, these results suggest that the TGF-α/EGFR pathway is required for the increased endothelial branching induced by LM8-hPHD4 cells.
TGF-α expression in PHD4-overexpressing LM8 tumors increases vessel density in vivo
Next, we examined the hypothesis that TGF-α stimulates angiogenesis in PHD4-overexpressing tumors in vivo. TGF-α–silenced LM8-PHD4 clones and a control clone with scrambled shRNA were injected into mice. No significant difference in tumor growth was observed between the TGF-α–silenced and the control clones (Fig. 6A). However, histologic examination of tissue sections revealed that TGF-α–silenced tumors displayed significantly reduced vessel density (Fig. 6B). These results demonstrate that TGF-α stimulates angiogenesis in PHD4-overexpressing tumors, but does not influence tumor growth.
Knockdown of TGF-α in LM8-hPHD4 cells reduces vessel density but does not alter tumor growth, whereas knockdown of HIF-2α in LM8 cells inhibits tumor growth without affecting vessel density. A, control LM8-hPHD4 cells (hPHD4#11-shscr9) and LM8-hPHD4 cells silenced for TGF-α were transplanted into mice, and tumor growth was monitored over a time period of 21 days. B, subsequently, tumors were prepared, sections were stained for PECAM, and vessel density was determined as described in Materials and Methods. Each bar represents the mean ± SEM of 5 analyzed tumors. C, control LM8-pcDNA3-shscr1 cells (which express HIF-2α) and LM8-pcDNA3 cells silenced for HIF-2α (clones LM8-pcDNA3-shHIF2α9 and LM8-pcDNA3-shHIF2α19) were transplanted into mice, and tumor growth was monitored over a time period of 20 days. D, subsequently, tumors were prepared and vessel density was determined as for LM8-hPHD4shTGF cells. Each bar represents the mean ± SEM of 5 analyzed tumors. Asterisks indicate significant (*, P < 0.05; **, P < 0.01; and ***, P < 0.001) differences.
HIF-2α silencing inhibits LM8 tumor growth but does not alter vessel density in vivo
The observation that TGF-α does not influence tumor growth raised the question after the mechanisms by which PHD4 overexpression inhibits tumor growth in vivo. Because HIF-2α was shown to have prooncogenic activity in various tumors and was downregulated in LM8-PHD4 cells, we suspected that silencing of HIF-2α in LM8 cells might inhibit tumor growth. LM8-pcDNA cells were stably transduced with shRNA specific for HIF-2α. LM8-pcDNA-shHIF2α clones 9 and 19 showed the strongest reduction of HIF-2α (Supplementary Fig. S5) and were tested in tumor experiments. Tumor growth of these cells was strikingly inhibited in comparison with LM8-shscr control tumors (Fig. 6C). However, there was no difference in vessel density between the HIF-2α–deficient and control tumors (Fig. 6D).
Discussion
Hypoxia is one of the most critical environmental factors involved in the progression of solid tumors. It contributes significantly to the angiogenic switch that converts a dormant tumor into an invasive, rapidly growing tumor (42–44). Less well studied, but also of critical importance for tumor pathophysiology, is the influence of hypoxia on glucose metabolism, pH regulation, tumor cell proliferation, apoptosis, and survival. Tumor hypoxia can lead to the development of an aggressive tumor cell phenotype and has been suggested as an adverse prognostic factor for patient outcome (45, 46). Clinical data show an association between elevated HIF-1α or HIF-2α protein levels and increased patient mortality in various human cancers, consistent with the observation that HIF-1 and HIF-2 regulate the expression of a large number of genes with important functions in tumor biology (2). Therefore, it has been proposed that manipulating oxygen-sensitive signaling pathways might be a promising strategy for treating malignancies (15). In support of this hypothesis, various studies have shown that the loss of HIF-1α or HIF-2α function can inhibit the growth of experimental tumors (reviewed by Semenza in ref. 2). However, this correlation is not absolute. In certain tumor models, HIF-1α or HIF-2α deficiency can also inhibit tumor cell apoptosis, thereby enhancing tumor growth (16, 18, 47). This discrepancy reflects the complexity of the cellular response to hypoxia, which is due to the large number of genes regulated by HIF-1 and HIF-2 (2). In addition, cell type- and tumor type–specific differences in the hypoxia response may exist. Whether the combined HIF-1 and HIF-2 inhibition in tumor and stromal cells will have the highest therapeutic efficacy, remains to be determined.
A potential alternative to directly targeting the HIF pathway could be to upregulate the expression of PHDs, which function as negative regulators of HIF-1α or HIF-2α. Consistent with this hypothesis, the overexpression of PHD1 in colon carcinoma cells suppressed HIF-1α accumulation and secretion of VEGF, leading to reduced vascularization and growth of experimental tumors (31). Conversely, PHD2 silencing stimulated tumor vascularization (29, 32), demonstrating that tumor angiogenesis is also limited by PHD2. Unexpectedly, however, PHD2 acted in a HIF-independent manner in the tumor models studied. The existence of various non-HIF targets (28, 48) implies that the function of PHDs is not restricted to the regulation of the cellular hypoxia response and that interfering with PHD proteins in tumors might have effects that are different from those resulting from HIF inhibition.
Here, we tested the hypothesis that experimental PHD4 overexpression might similarly block tumor growth and angiogenesis in a syngeneic mouse osteosarcoma model. A key finding of our study is that tumor growth was inhibited, whereas tumor vessel density was significantly enhanced. This result is provocative in that solid tumor growth is thought to be dependent on the expansion of the host vasculature (42, 49). However, the newly formed vessels in PHD4 overexpressing tumors were small and partially nonfunctional, as indicated by impaired tissue perfusion. Thus, PHD4 downregulation in tumor cells might be a mechanism that restricts excessive and nonproductive angiogenesis. That experimental tumor growth can be inhibited together with an increase in vascularization is not without precedent: blockade of the Notch ligand, Dll4, was shown to inhibit tumor growth by stimulating nonproductive vascularization, providing a striking example of the uncoupling of angiogenesis and tumor growth (50). However, nonfunctional angiogenesis was not the cause of LM8-hPHD4 tumor growth inhibition, as evidenced by the results of the TGF-α knockdown experiments. Rather, tumor cell intrinsic mechanisms inhibit their proliferation in vivo. We observed that the knockdown of HIF-2α in LM8 cells greatly blocks tumor growth, suggesting that the reduction of HIF-2α expression in LM8-PHD4 cells is the cause of tumor growth inhibition.
Our observation that PHD4 overexpression in tumor cells caused excessive tumor vascularization distinguishes this prolyl hydroxylase from PHD1 (31) and PHD2 (29, 32), which inhibit tumor angiogenesis. Given the differences in their expression patterns in tissues and in their subcellular localization, it is likely that the individual members of the PHD family influence different signaling pathways including angiogenic pathways. Unlike PHD1, whose activity inhibited VEGF production by tumor cells via suppression of HIF-1α, PHD4 did not influence VEGF expression, presumably because HIF-1α levels were unaffected. Yet, PHD4 overexpression in LM8 tumor cells resulted in a robust upregulation of TGF-α. This growth factor is known to stimulate angiogenesis via EGFR in endothelial cells (40, 41). Our results show that the TGF-α/EGFR pathway is required to stimulate angiogenesis induced by PHD4 overexpression, and is sufficient to stimulate angiogenesis in vitro in the context of LM8 cells. Although it is likely that TGF-α acts directly on endothelial cells, it cannot be excluded that its effect on vascularization is indirect and results from autocrine activity on tumor cells in vivo. Further studies are also needed to determine how TGF-α expression is upregulated by PHD4 in LM8 cells.
In conclusion, we show that PHD4 overexpresssion in the tumor cell compartment has a profound effect on vessel density, which is caused by TGF-α upregulation. The influence of PHD4 on tumor angiogenesis is fundamentally different from the effects exerted by other PHDs. Whereas PHD1 and PHD2 inhibit neovascularization, PHD4 stimulates this process, showing that the different members of the PHD family control different angiogenic pathways. Taken together, our results lend further support to the hypothesis that PHDs regulate the angiogenic balance in tumors. Yet, our observations and those made by others (50) challenge the traditional view that high tumor vessel density necessarily correlates with increased perfusion and enhanced tumor growth. Moreover, our results show that HIF-2α, although expressed at relatively low levels, has prooncogenic activity in LM8 allografts, and suggest that PHD4 overexpression inhibits tumor growth by reducing HIF-2α expression. Despite the complexity of the cellular pathways regulated by PHDs, manipulating of their expression may turn out to be a useful approach for therapeutic intervention in tumors.
Disclosure of Potential Conflicts of Interest
I. Flamme is employed with Bayer Schering Pharma. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: A. Klotzsche-von Ameln, I. Prade, I. Flamme, B. Wielockx, G. Breier
Development of methodology: A. Klotzsche-von Ameln, I. Prade, M. Rezaei, I. Flamme, B. Wielockx, G. Breier
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Klotzsche-von Ameln, I. Prade, M. Grosser
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Klotzsche-von Ameln, I. Prade, M. Grosser, A. Kettelhake, M. Rezaei, B. Wielockx, G. Breier
Writing, review, and/or revision of the manuscript: A. Klotzsche-von Ameln, I. Prade, T. Chavakis, I. Flamme, B. Wielockx, G. Breier
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Klotzsche-von Ameln, I. Prade, A. Kettelhake, B. Wielockx, G. Breier
Study supervision: I. Prade, B. Wielockx, G. Breier
Grant Support
This work was supported in part by grants of the Deutsche Forschungsgemeinschaft (DFG-Br 1336/2-2 and 2-3, DFG-Wi 3291/1-1), the Bundesministerium für Bildung und Forschung (to G. Breier and A. Klotzsche-von Ameln), and the Daimler-Benz-Foundation (to G. Breier) and European Research Council (ENDHOMRET to T. Chavakis).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Acknowledgments
The authors thank Drs. Christian Beltinger (University Children's Hospital Ulm, Germany) and Antonis Hatzopoulos (Vanderbilt University, Nashville, TN) for providing LM8 cells and eEPCs, respectively. The expert technical assistance of Anke Klawitter, Antje Muschter, Rosi Müller, and Annelie Zürich is gratefully acknowledged.
Footnotes
Note: Supplementary data for this article are available at Molecular Cancer Research Online (http://mcr.aacrjournals.org/).
- Received April 23, 2013.
- Revision received August 5, 2013.
- Accepted August 19, 2013.
- ©2013 American Association for Cancer Research.
References
Volume 11, Issue 11
