Angiopoietin-2 has been implicated in the angiogenic response; however, this response has been tied to the expression of VEGF, and an independent angiogenic role has yet to be described. In this report, we detail the generation of transgenic mice that conditionally express angiopoietin-2 in the liver, resulting in sustained increases in circulating levels. These animals survive gestation and present with several vascular abnormalities, including an increase in the diameter of myocardial coronary vessels and a reduction in the density of endocardial vessels. In the lung, prominent increases in vessel diameter were observed. These vascular remodeling changes occurred in the absence of any apparent increase in VEGF expression. Our results illustrate that chronic systemic delivery of angiopoietin-2 induces angiogenesis in the absence of increased VEGF expression and that angiopoietin-2 promotes myocardial coronary vessel remodeling.
vascular development is tightly regulated by various factors, including VEGF and the angiopoietins. The angiopoietin (Ang) family is composed of four members, Ang-1 to -4 (24); Ang-3 (mouse) and Ang-4 (human) are interspecies orthologs. Original studies on Ang-3 and Ang-4 suggested that Ang-4 was an agonist, whereas Ang-3 was an antagonist for Tie-2 activation (24). However, it would now seem that both ligands are agonists and that the original difference noted was probably due to species-dependent effects (11). The first member of the Ang family, Ang-1, has been the most widely studied. Ang-1 appears to work in complementary fashion with VEGF-A during early vascular development, with VEGF-A initiating vascular formation (2, 5) and Ang-1 promoting subsequent vascular remodeling, maturation, and stabilization (3, 20).
Ang-2 has been the subject of many studies, but its role is still controversial. Early studies using human umbilical vein endothelial cells suggested that Ang-2 inhibited Ang-1-mediated autophosphorylation of Tie-2 (14). Other in vitro studies (14, 16, 22) have demonstrated an agonistic role for Ang-2. Recently, a number of studies (6, 14) have focused on the role of Ang-2 in vascular development in vivo by using mouse molecular genetic approaches. One study described the effect of overexpression of Ang-2 during embryo development (14), and a second studied the effect of the deletion of Ang-2 on vascular development (6). Overexpression of Ang-2 during embryonic development was shown to be lethal as a result of induction of widespread vessel discontinuities (14), presumably by antagonizing Ang-1 through the Tie-2 receptor tyrosine kinase. According to these studies, Ang-2, unlike VEGF and Ang-1, is not required during embryonic vascular development. Mice homozygous for the altered allele (Ang-2-/-) were born at normal frequencies with defects in lymphatic development. However, the majority (up to 95%) of these mice died within 2 wk of birth, suggesting a role for Ang-2 in postnatal lymphangiogenesis.
In adults, Ang-2 expression is readily detectable only in the ovary, placenta, and uterus (14), which are the three predominant sites of vascular remodeling in the normal adult. Postnatal angiogenesis is associated with numerous pathological conditions, including atherosclerosis, arthritis, retinopathy, and neoplastic tumor growth. Recent studies have shown that plasma Ang-2 levels are increased in patients with acute coronary syndrome (12) and that Ang-2 expression was upregulated in ischemic myocardium (7, 15). Moreover, expressions of Ang-2 and Tie-2, but not Ang-1, were also increased in angiosarcoma and acquired immunodeficiency syndrome-associated Kaposi’s sarcoma (1) as well as in rodent tumor tissues (8). Ectopic expression of Ang-2 in tumor cells induced angiogenesis in association with increased tumor growth in transplanted mice (4, 21). Furthermore, a role for Ang-2 in tumor-mediated angiogenesis was suggested in xenograft experiments, where selective inhibition of Ang-2 resulted in inhibition of tumor cell growth (17).
These differences in the response of the vasculature to the angiopoietins may be very dependent on the microenvironment, or context, in which the endothelial cell is found. The aim of this study was to examine the effect of Ang-2 delivery to different vascular beds. For this, we generated mice that conditionally expressed human angiopoietin-2 (hAng-2) in a hepatocyte-specific manner, therefore, secreting hAng-2 into the systemic circulation. We demonstrate that the transgenic hAng-2 is delivered in the circulation, and, quite surprisingly, for the first time, we demonstrate that hAng-2 has angiogenic activity that does not seem to require the upregulation of the potent angiogenic factor VEGF.
MATERIALS AND METHODS
Transgene expression analyses of transgenic mice.
The plasmid-expressing tetracycline (pTet)-hAng-2 responder was generated by oocyte injection of the hAng-2 construct, as previously described (23), and the liver-enriched activator protein (LAP) driver lines have been described previously (10, 18, 23). Crosses were performed between pTet-Ang-2 and the LAP tetracycline-responsive transactivator (tTA) lines from CD-1 background, and offspring were genotyped by PCR by using DNA extracted from tail biopsies. DNA was prepared and PCR performed by using primers as previously described (18). Single and wild-type (WT) littermates served as experimental controls.
Transgene expression was determined by using RT-PCR analyses. Liver RNA was isolated with the use of TRIzol (Invitrogen, Carlsbad, CA), treated with DNase (Sigma-Aldrich, St. Louis, MO), and subjected to two-step RT-PCR (Clontech, Palo Alto, CA), all according to each manufacturer’s suggested protocol. Samples were run in duplicate, with one set undergoing RT and a second set containing no RT to control for possible DNA contamination. PCR was completed on cDNA by using the following primers for β-actin and (transgene-specific) hAng-2: 5′β-actin-primer-GTGGGCCGCTCTAGGCAC-CAA; 3′β-actin-primer-CTCTTTGATCTCACGCACGATTTC; 5′-hAng-2-primer-GCCTGGAGAGAACACAGCAG; and 3′-hAng-2-primer-AA-CTTG-AGGGCAAACACACG.
The secretion of hAng-2 into the systemic circulation was confirmed by performing an ELISA specific for hAng-2 in plasma samples, according to the manufacturer’s instruction (BD Biosciences Clontech).
Frozen sections of lung and heart tissues were homogenized in radioimmunoprecipitation assay buffer and quantified by using the bicinchoninic acid protein assay (Pierce, Rockford, IL). Equal loading of samples was confirmed by performing Western blotting on 30 μg of total protein from tissue lysates, as described previously (9), by using a monoclonal antibody against β-actin (Clone AC-15, Sigma) at a 1:5,000 dilution and an anti-mouse horseradish peroxidase-conjugated secondary antibody (1:5,000, Stressgen, Victoria, British Columbia, Canada). Tie-2 (also known as Tek) phosphorylation and expression levels were determined by immunoprecipitating 1 mg of protein with 2 μg of polyclonal anti-Tie-2 antibody (Santa Cruz Biotechnology, Santa Cruz, CA) as described previously (9). Immunoprecipitation was followed by Western blotting with the anti-phosphotyrosine antibody 4G10 (1:10,000; Upstate Biotechnology, Lake Placid, NY) and an anti-mouse horseradish peroxidase-conjugated secondary antibody as described above. Blots were stripped and reprobed with a mouse monoclonal anti-Tie-2 antibody (1:5,000; Upstate Biotechnology) and an anti-mouse horseradish peroxidase-conjugated secondary antibody as described above. Blots were reproduced four (4) times.
All animals received humane care according to the Canadian Council On Animal Care Guidelines. The study was reviewed and approved by the Sunnybrook Research Institute Animal Care Committee based on the guidelines provided by the Canadian Council on Animal Care. Animals were kept on a standard light cycle and had ad libitum access to food and water throughout the course of the experiments.
A total of 26 double-transgenic (DT) mice versus 25 WT mice were exposed to hAng-2 driven from the transgene from conception to adult life by not adding doxycycline (Dox) in their drinking water. Mice were provided with deep anesthesia by inhalation of isoflurane in 100% O2. Blood samples for the ELISA test were taken by cardiac puncture, and the organs were removed and immediately observed and photographed under a dissecting microscope with the use of a digital camera (PixelInk). The organs were then dissected into different portions for histology, protein, and RNA assays.
Fluorescent microangiography imaging of lung and heart.
Fluorescent microangiography was conducted on DT and WT mice as previously described (Ref. 6a). The mice [10-wk-old DT (n = 5) and WT (n = 5) littermates] were anesthetized with ketamine and xylazine. A heparin bolus (50 U) was injected via the right ventricle after a thoracotomy was performed. Littermates were perfused transcardially via the right and left ventricles with PBS containing heparin (1 U/ml), followed by an injection of 1:10 (vol/vol) of yellow-green fluorescent microspheres (0.2-μm size) (Molecular Probes, Invitrogen, Burlington, Ontario, Canada) suspended in 1% (wt/vol) low-melting agarose gel (Sigma-Aldrich). The perfusion (1.5 ml of low-melting agarose solution suspended with 0.2-um size fluorescent microbeads) was injected via the right ventricle while the lungs were being inflated by a tracheal tube. Copious amounts of wet ice were dumped onto the chest wall when the fluorescent solution oozed out from the incised left atrium. The LV infusion approach was used for imaging other organs, such as the heart, liver, spleen, and kidneys.
The hearts and the lungs were dissected out and fixed in 4% paraformaldehyde in PBS for 48 h at 4°C and then transferred in PBS for 48 h. Each organ was sectioned (150-μm sections) by using a vibratome (Leica VT1000S) and counterstained by propidium iodide (10 μg/ml; Sigma-Aldrich) in PBS for 10 min. Sections were mounted on a slide with antifading mounting medium (Vector, Burlingame, CA) and observed under a confocal microscope (Bio-Rad). A serial scan was performed for each section, and scans from the DT mice were compared with those of the WT mice.
Quantitation of vessels in fluorescent microangiographic images.
The IPTK 4.0 imaging software (Image Processing Toolkit, Reindeer Graphics, Asheville, NC) was used to quantify vessels in fluorescent microangiographic images of lungs. The vessel score/vascular density for each image was presented as the computed value of ends + nodes/2.
The lungs and liver were fixed overnight in 10% buffered formalin, embedded in paraffin, cut in 5-μm sections with the use of a microtome, and stained with Masson Trichrome. Immunohistochemistry of the gut was performed by using a hamster polyclonal anti-podoplanine (1:200) generated in our laboratory. Antigen was detected through the use of a goat anti-hamster secondary antibody at 1:150 (Vector) and diaminobenzidine peroxidase substrate (Vector). Slides were counterstained with methyl green, dehydrated in ethanol, and mounted with permanent mounting medium.
Immunohistochemistry of the lung and liver was performed by using (1:500) rat anti-mouse platelet endothelial cell adhesion molecule (PECAM) (Pharmingen, San Diego, CA). Antigen was detected through the use of rabbit anti-rat secondary antibody at 1:150 (Vector). Slides were counterstained with hematoxylin.
Mouse VEGF level was assessed by performing an ELISA specific for mouse VEGF in plasma samples, according to the manufacturer’s instruction (BD Bioscience Clontech).
Systemic delivery of Ang-2.
The complex nature of vascular remodeling is known to involve the choreography of the expression of many different angiogenic factors, including Ang-2. The role of Ang-2 in the angiogenic response remains unclear; thus, in an attempt to gain greater insight into the role of this factor in vessel remodeling, we expressed hAng-2 in a hepatocyte-specific manner using a Dox-based binary transgenic expression system, as described previously (23). Two independent lines of mice that responded well in the presence of the driver line were chosen for these studies, and there were no marked differences in this phenotype (data not shown), demonstrating that the derived phenotype was dependent on the expression of hAng-2 and not on the integration event. The use of this binary transgenic system has been described previously (23): briefly, the tTA expressed from the LAP promoter (driver transgene) binds to the tetracycline operator sequences upstream of the hAng-2 cDNA (responder transgene). In the absence of the tetracycline analog Dox, Ang-2 is expressed; in the presence of Dox, expression of hAng-2 is suppressed.
DT mice survive gestation at the estimated Mendelian ratio with no statistically significant difference in size and weight when compared with their control littermates (data not shown). RT-PCR analyses with the use of specific primers recognizing human and not mouse Ang-2 confirmed the expression of transgene-derived hAng-2 in the liver of DT animals and not in WT mice (Fig. 1A). An ELISA test specific for hAng-2 on the plasma confirmed the presence of hAng-2 in the blood of DT mice. ELISAs were performed on all the DT mice and showed variable amounts of hAng-2 secreted (Fig. 1B), with a proportion of the mice (46%) having no detectable amount of hAng-2 in their blood, although the transgene was expressed, albeit at very low levels, in these animals. There was a very good correlation of transgene expression levels and the amount of hAng-2 found in the blood (data not shown).
Increased vascular density in hAng-2-expressing animals.
Antemortem examination of DT mice did not reveal any pathological symptoms compared with their WT littermates, except for a reddish-bluish appearance of the mucosa of the mouth and nose (Fig. 2A), foot pads (Fig. 2B), and skin, most apparent in the ears (Fig. 2C). No mortality occurred during the adulthood of the mice. The examination of the organs after the mice were killed indicated gross abnormalities in DT organs, mainly in the lungs, heart, and liver. The liver presented with uncommon multiple vessels in its capsule (Fig. 2E). The heart was enlarged and presented with a dilatation of the coronary vessel (Fig. 2F), whereas the lungs were congested (Fig. 2D). None of these gross abnormalities was observed in the organs of WT mice.
Histological examination of those organs provided further support for these macroobservations. Masson Trichrome staining of liver sections showed that DT mice had disorganized and enlarged blood vessels (Fig. 3A) compared with their WT littermates (Fig. 3B). PECAM immunostaining of sections from DT livers further demonstrated increased size of blood vessels in DT mice (Fig. 3C) compared with WT mice (Fig. 3D). In the lungs, a similar increase in blood vessel size was identified in DT mice compared with WT mice (Fig. 4). These changes were similar at all ages sampled, beginning as early as 4 wk of age and extending up to 24 wk. Somewhat surprisingly, the severity of the phenotype did not seem to correlate well with the amount of secreted hAng-2 detected, which may reflect that even the low level detected is sufficient to drive the complete phenotype and that excess hAng-2 has no consequence. Importantly, none of these changes was observed in the normal animals or in DT mice, where we were unable to detect the secretion of hAng-2.
To better visualize the vasculature, we performed fluorescent angiography specific to the arteries. This put in evidence an increase in vessel size in the lungs of DT mice (Fig. 5A) compared with WT mice (Fig. 5B), supporting the histological observations. Furthermore, this difference was quantified by using dedicated vessel imaging software (Table 1). The vessel score for each image was presented as the computed value of ends + nodes/2. This analysis clearly demonstrated that the vessel score density values were much higher in DT mice (5,146 and 5,368) compared with WT mice (2,878 and 2,024). This probably contributed to the dramatic congestion observed in the lungs. Importantly, this approach allowed us to visualize a remarkable increase in arterial density in the lung, something that was not readily apparent in the histological sections. In contrast to the lung, the vasculature of the endocardium was apparently reduced, whereas the coronary vessels in the myocardium appeared enlarged (Fig. 5C) compared with WT mice (Fig. 5D).
hAng-2-expressing mice have increased lymphatic vessel patterning.
To determine whether the production of hAng-2 influenced the lymphatic system, cross sections of the intestine, a tissue with a well-defined lymphatic system, were specifically labeled for lymphatic vessels by using anti-podoplanine antibody. An increase in lymphatic vessel patterning was observed in the DT mice (Fig. 6A) compared with WT mice (Fig. 6B).
Tie-2 phosphorylation is not reduced in hAng-2-expressing mice.
Ang-2 is known to be a context-dependent agonist of Tie-2 activity; thus we set out to examine the activation state of Tie-2 from Ang-2-expressing animals. Whole tissue lysates from the lung and the heart were analyzed for Tek expression and tyrosine phosphorylation by using anti-Tek and anti-phosphotyrosine antibodies, respectively. The levels of Tie-2 protein and its phosphorylation status did not change between DT and WT mice (Fig. 7).
VEGF level is unaltered in hAng-2-expressing mice.
Ang-2 has previously been shown to work in concert with VEGF to induce vascular growth; thus we investigated the expression of VEGF in hAng-2-expressing mice. VEGF levels in the plasma did not vary between DT and WT mice (2.5 vs. 2.3 pg/ml, respectively), suggesting that the angiogenic changes in these animals are not driven by the elevation of VEGF levels.
In this study, we have utilized a liver-specific binary transgenic system to deliver hAng-2 systemically over a chronic period to examine the role of Ang-2 on different vascular beds during development and in the adult. Although many vascular and lymphatic abnormalities were observed, none was severe enough to produce embryonic lethality. The penetrance of the vascular phenotype did vary from unaffected to severe, which is probably a reflection of the random-bred CD-1 background from which these animals were constructed. Nonetheless, the vascular abnormalities in the lung and heart appeared with the greatest consistency and as such were the focus of this study. The coronary vessels of the heart exhibited extensive dilatation and increased tortuosity, accompanied by a decrease in endocardial vessel density. The vasculature of the lung was increased in size. Histological analysis and fluorescent angiography revealed an increase in both the size and the complexity of the blood and lymphatic vasculatures.
Early studies (14) that utilized Tie-2-promoter-driven-Ang-2 transgenic animals demonstrated that Ang-2 was not compatible with early development. These experiments employed a transient transgenic approach, which did not permit the analysis of numerous animals; however, it was apparent with the limited number of animals analyzed that some Tie-2-Ang-2 transgenic embryos died in utero at embryonic (E) day 9.0 (E9.0), with vascular defects that resembled those of the Tie-2 knockout embryo (14). These studies provided the initial foundation for the Tie-2/Ang-signaling paradigm that suggested that Ang-1 was the agonistic ligand, whereas Ang-2 antagonized the activity of Tie-2. Although several studies have supported this paradigm, several other studies have suggested that Ang-2 does have agonistic properties that are dependent on the microenvironment of the endothelial cell, i.e., “context dependent.”
In contrast, our study demonstrated that the expression of Ang-2 in the early hepatocyte is not lethal and is well tolerated. This discrepancy most probably is due to the timing of the expression and the cell type that expresses the transgene. The expression of Tie-2 is known to begin as early as E7.0 in the early vascular progenitors (19), whereas the expression of the LAP-tTA-driver transgene is first detected in the liver primordial at E9.5 (19). Another possibility for the difference in the tolerability of Ang-2 in these animals is that Ang-2 delivered in a paracrine fashion is better tolerated than when produced in an autocrine fashion.
Overexpression of hAng-2 in hepatocytes provided an efficient means to deliver this factor into the systemic circulation. In this study, we show that hAng-2 delivered in a chronic fashion in the circulation can affect the vascular beds of several organs. These effects extend to both the blood and lymphatic vessels. A role for Ang-2 in lymphangiogenesis has been suggested by mouse molecular studies, where a knockout of the Ang-2 gene resulted in Ang-2-null mice that survived gestation but died postnatally. These animals displayed disorganization and hypoplasia of the lymphatic capillaries (6). The disorganization of lymphatic microvessels in our study further supports a role for Ang-2-mediated signaling in lymphatic development.
Ang-2 is thought to collaborate with VEGF to stimulate vessel growth by antagonizing Tie-2 activity, resulting in the destabilization of the vessels and allowing these vessels to respond to VEGF action. Although we and others have shown that Ang-2 has angiogenic activity in vitro, until now a direct in vivo angiogenic role for Ang-2 has not been reported. In this study, we show that hAng-2 promotes angiogenesis in the lung and the heart without a concomitant increase in VEGF production, suggesting that Ang-2 alone may, in fact, have independent angiogenic activity. Nonetheless, it is possible that the high level of Ang-2 in the circulation had contributed to destabilization of vessel walls and consequently opened the way for VEGF action. The elevated level of Ang-2 may also reduce the threshold level for VEGF-R2 activation and signaling.
The examination of a cross section of the heart after fluorescent angiography in the mice producing an excess amount of Ang-2 in their circulation showed a dramatic dilatation of coronary vessels and a reduction of vasculature in the endocardium. The involvement of Ang-2 in coronary angiogenesis along with VEGF has been previously documented (13, 15). According to Lee et al. (12), the plasma levels of Ang-2 increased in patients with acute coronary syndrome. Furthermore, an upregulation of both Ang-2 and VEGF expression was also described in ischemic myocardium (15), with the patients who showed evidence of myocardial damage having the highest levels of Ang-2 (12). However, in all these studies, no direct evidence has been demonstrated that increased plasma levels of Ang-2 reflect the angiogenesis activity in the ischemic myocardium. Here we show for the first time that Ang-2 may have a direct effect on coronary vessel remodeling and that this activity is not dependent on the upregulation of VEGF.
The hypovascularization of the endocardium observed in our study in response to Ang-2 in the circulation could be due to a change in the association between the endocardium and underlying myocardium. A somewhat similar, although more severe, defect has been described in early mouse embryos lacking either Ang-1 or Tie-2 (14). Thus it may be possible that the endocardial endothelial cells respond to Ang-2 as an antagonist of Tie-2 signaling within this context.
In the lungs, an increase in the number and the size of blood vessels of DT mice compared with WT mice was detected after perfusion of the arteries. However, histological sections from the lungs of these mice did reveal an increase in the size of blood vessels but not in their total number. The mucosa of DT mice had a bluish appearance, suggesting cyanotic conditions that might be associated with arterio-venous shunting, and this might explain the increase of the number of blood vessels observed in DT mice when performing fluorescent microangiography of the arteries.
Until now, it was believed that Ang-2 required the collaboration of VEGF to promote angiogenesis. In the present study, we demonstrate for the first time that a direct action of Ang-2 in promoting angiogenesis can occur without the upregulation of VEGF expression. We also show that circulating Ang-2 does not result in a either a decrease or an increase in Tie-2 phosphorylation; thus Ang-2 does not appear to be acting as an antagonist. However, it still remains unclear how Ang-2 exerts its effect. The antiphophotyrosine antibodies used in these studies are not directed to specific sites but to pan-phosphotyrosine; thus, if there were changes to the repertoire of sites phosphorylated on Tie-2 in response to Ang-2 that did not change the total amount of tyrosine phosphorylation, it would not change the overall signal intensity, which does not allow us to discern changes at specific sites in Tie-2. Nevertheless, our data point to the fact that Ang-2 has angiogenic activity in the absence of elevated VEGF, a point that is further supported by a recent study (17) that specifically targeted Ang-2 in tumor development and found it to be antiangiogenic.
This work was supported by the National Cancer Institute of Canada.
We thank Maribelle Cruz for help with the animal husbandry and Sue Santillo for excellent administrative support. The authors also thank Michael Kuliszewki for helping with vessel quantitation of fluorescent microangiographic images.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 by the American Physiological Society