We examined the RNA content of the gene encoding angiopoietin (Ang)-2, a modifier of angiogenesis, in hepatic metastases of colorectal cancer (CRC) to explore the role of this protein in neovascularization of metastatic foci. Metastatic CRC exhibited notable blood flow and tumor vessel formation at tumor frontiers. Reverse-transcription polymerase chain reaction assays indicated that the ANG2 RNA content was greater in metastatic CRC than in primary CRC. Investigation of metastatic foci using laser capture microdissection revealed that the RNA content of ANG2, but not ANG1, increased from the bordering liver region to the periphery of the metastatic disease, and also from the periphery to the intermediate portion of the metastatic lesion; immunohistochemical analysis confirmed that there was a corresponding gradual increase in Ang-2 protein expression. Tie-2, a receptor for angiopoietins, was preferentially expressed in the bordering liver region rather than in metastatic CRC. Vascular endothelial growth factor (VEGF) also exhibited an expression pattern similar to that of Ang-2, and there was a significant correlation between the RNA content of ANG2 and that of VEGF in dissected samples (P = .002). Western blot analysis suggested that expression of Ang-1, Ang-2, Tie-2, and VEGF may be regulated at a transcriptional level. The increase in ANG2 RNA content from the peripheral portion of the tumor to the intermediate portion, coinciding with the decrease in recruitment of periendothelial supporting cells around the vascular endothelial cells, suggests that Ang-2 may play a role in the immaturity of tumor vessels. In conclusion, the current study suggests that Ang-2 and VEGF may cooperate to enhance the formation of new blood vessels in metastases of CRC to the liver. (HEPATOLOGY 2004;39:528–539.)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Colorectal cancer (CRC) is a common malignancy worldwide. Although 84%–92% of patients with CRC are treated with surgical resection, more than half of these patients subsequently develop disease recurrence,1 the most common type of recurrence being CRC metastatic to the liver, which often is associated with mortality.2, 3 Metastasis of CRC to the liver is a complex multistep process, which includes adherence of metastatic cells to endothelial cells (ECs), invasion across the endothelial basement membrane, cell proliferation, and neovascularization.4 It was reported that tumor vessels appeared in human liver metastases when metastatic foci grew to 200 μm in diameter and that the density of tumor vessels increased as tumor size increased.5 Angiogenesis is essential for tumor growth and expansion, because the resulting blood vessels supply malignant cells with sufficient oxygen and nutrition.6, 7 Therefore, interruption of this process is considered to be a strategy for preventing CRC metastases to the liver.
Recent studies have focused on novel endothelial growth factors such as angiopoietins (Ang), which are ligands for the endothelium-specific tyrosine kinase receptor Tie-2.8–10 Ang molecules play crucial roles in normal vascular development and in embryonic angiogenesis. Of the four currently known Ang molecules (Ang-1, Ang-2, Ang-3, and Ang-4), the best-characterized ones are Ang-1 and its natural antagonist, Ang-2. Ang-1 is widely expressed in normal adult tissue,11 whereas Ang-2 is expressed primarily at sites of vascular remodeling, such as the ovaries, uterus, and placenta.10 Angiogenesis requires migration and remodeling of ECs derived from preexisting blood vessels and regulation of the perivascular microenvironment. Ang-2 destabilizes preexisting vessels by weakening interactions between ECs and periendothelial supporting cells (PESCs).10, 12 Ang-1 subsequently acts, via the Tie-2 receptor, to remodel these primitive vessels and to help maintain and stabilize mature vessels.9, 13
Ang-2 is expressed in several types of human malignancies, including carcinomas of the colon, liver, stomach, lung, and thyroid, as well as malignant glioma.14–21 Furthermore, gene transduction studies have demonstrated that forced expression of the ANG2 gene in gastric and colon carcinoma cells is associated with increased vessel density.16, 22 These findings suggest that Ang-2 may be involved in tumor-associated neovascularization. Nonetheless, expression of Ang-2 in CRC metastases in the liver has not been fully characterized.
Although CRC metastatic to the liver is considered to be a hypovascular lesion (in contrast to hypervascular hepatocellular carcinoma), it appears that the periphery of the metastatic lesion retains rich vascularity.23, 24 We have examined the expression of ANG2 RNA in CRC metastases in the liver and compared the results with those observed in primary CRC tissue samples using the reverse-transcription polymerase chain reaction (RT-PCR) assay. Using laser capture microdissection (LCM), RNA content data on ANG2, ANG1, TIE2, and the VEGF gene, which encodes the putative angiogenic factor VEGF (vascular endothelial growth factor), were obtained for normal liver cells and for metastatic CRC cells in the liver, with reference to vessel density in each region of the liver. Our results represent notable findings regarding tumor-associated neovascularization in CRC metastases in the liver and the possible link between this neovascularization and the concurrent expression of Ang-2 and VEGF.
Surgical specimens were obtained from patients who had undergone colectomy and/or hepatectomy at the Department of Surgery and Clinical Oncology at Osaka University (Osaka, Japan) between 1995 and 2001 because of CRC (n = 36) and/or CRC metastatic to the liver (n = 14; 8 solitary lesions and 6 multiple lesions). Nine liver metastases were found simultaneously with primary CRC (synchronous lesions), and 5 liver metastases were found at least 1 year after detection of primary CRC (metachronous lesions). Metastatic CRC lesions ranged in size from 1.0 to 7.0 cm (mean ± standard deviation, 3.1 ± 1.9 cm). None of the patients had been treated preoperatively with chemotherapy or radiotherapy. All resected surgical specimens were fixed with 10% buffered formalin. A portion of each tissue sample promptly was frozen in liquid nitrogen for the RT-PCR assay or for Western blot analysis. Of the 14 CRCs metastases of the liver, 10 were determined to be of sufficient quantity and were made into optimal cutting temperature compound–frozen samples.
Total RNA was extracted from clinical samples and cell pellets via a single-step method using Trizol reagent (Life Technologies, Gaithersburg, MD), and complementary DNA (cDNA) was generated using avian myeloblastosis virus reverse transcriptase (Promega, Madison, WI) as described previously.25
Frozen tissue samples embedded in optimal cutting temperature compound were sectioned at 8 μm, mounted on uncoated glass slides, fixed with 70% ethanol for 10 minutes, and then washed with distilled water for 30 seconds. Sections were stained rapidly with hematoxylin-eosin (HE) solution, washed twice with 95% ethanol and once with 100% ethanol, and then washed with xylene. After 20 minutes of air-drying, LCM was performed using the LM200 LCM system (Arcturus Engineering, Santa Clara, CA). Sections were covered with transfer film (CapSure TF-100; Arcturus Engineering), and the targeted cell population was dissected with the laser beam and captured onto the film. The RNeasy minikit (Qiagen, Hilden, Germany) was used to extract a minimal amount of RNA from each tissue sample, as described previously.26
Semiquantitative duplex. RT-PCR.
Semiquantitative analyses of the expression of ANG1, ANG2, TIE2, and VEGF RNA were performed using the duplex RT-PCR technique, as described previously.25, 27 β-actin was used as the internal standard. The sequences of the PCR primers and the sizes of the amplicons are shown in Table 1. PCR reactions were performed in a total volume of 25 μL, which consisted of 2 μL cDNA template, 1X Perkin-Elmer PCR buffer (Perkin-Elmer, Foster City, CA), 1.5 mM MgCl2, 0.8 mM deoxynucleotide triphosphates, appropriate amounts of β-actin and each primer, and 1 unit Taq DNA Polymerase (AmpliTaq Gold; Roche Molecular Systems, Branchburg, NJ). The amounts of each primer and β-actin primer used and the PCR conditions are shown in Table 2. PCR amplification was performed with a Genamp PCR System 9600 (Perkin-Elmer, Foster City, CA).
The anti-human antibodies used in the study were goat polyclonal Ang-1 antibody (Santa Cruz Biotechnology, Santa Cruz, CA), goat polyclonal Ang-2 antibody (Santa Cruz Biotechnology), rabbit polyclonal Tie-2 antibody (Santa Cruz Biotechnology), rabbit polyclonal VEGF antibody (Santa Cruz Biotechnology), mouse monoclonal CD31 antibody (Dako, Carpinteria, CA), mouse monoclonal α-SMA (smooth muscle actin) antibody (Dako), and rabbit polyclonal actin antibody (Sigma, St. Louis, MO). The blocking peptides that were used as immunogens for generation of the Ang-1 antibody and the Ang-2 antibody were obtained from Santa Cruz Biotechnology.
HE staining and immunohistochemistry.
Paraffin sections measuring 4 μm in thickness were deparaffinized in xylene and rehydrated and stained with HE solution for histopathologic examination. Immunohistochemical analysis of CD31, Ang-2, and VEGF was performed using the Vectastain avidin-biotin complex peroxidase kit (Vector Laboratories, Burlingame, CA), as described previously.25–27 Heat antigen retrieval was performed in 10 mM citrate buffer, pH 6.0, at 95 °C for 40 minutes. Primary antibodies were applied to sections at dilutions of 1:500 for CD31, 1:50 for Ang-2, and 1:50 for VEGF. As a positive control for Ang-2 staining, human placenta was used.10, 28 For the negative control, nonimmunized immunoglobulin G (Vector Laboratories) was used as a substitute for the primary antibody.
Blood vessels were counted with a microscope at 200× magnification after immunostaining of the vascular ECs with anti-CD31 antibody. Ten visual fields were randomly selected in each portion of the metastatic CRC lesion (i.e., the periphery, the intermediate region, and the center), and vessel counts per mm2 were calculated. For evaluation of vessel diameter in the shorter direction, more than 50 vessels were measured using Mac Scope Software (Mitani Corp., Fukui, Japan).27
Vessel maturation was defined as the extent of recruitment of PESCs around ECs and was expressed as the percentage of encirclement of vessel outline with α-SMA-expressing PESCs.10, 13, 16, 21 Thus, double-staining of ECs and PESCs was performed with anti-CD31 antibody and anti-α-SMA antibody, respectively. In brief, CD31 staining, which yields a brown color, was performed. After removal of the CD31 antibody by thorough washing in 0.1 M glycine solution, pH 2.2, for 1 hour, mouse monoclonal anti-human α-SMA antibody at a dilution of 1:200 was applied to the section for 2 hours at room temperature. This step was followed by incubation with anti-mouse secondary antibody conjugated with a dextran backbone containing alkaline phosphatase (EnVision AP; Dako) for 30 minutes. Color development (deep pink) based on alkaline phosphatase activity was achieved using fuchsin solution. For quantification, 10 blood vessels in each portion of the lesion were randomly selected under the microscope at 200× magnification and evaluated for maturation index.
Human colon cancer cell lines HCT116, SW480, and DLD1 were obtained from the American Type Culture Collection (Manassas, VA). These cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/mL penicillin, and 100 μg/mL streptomycin at 37 °C in a humidified incubator with 5% CO2 in air. Human umbilical vein ECs were grown on MCDB131 culture medium (Chlorella Inc., Tokyo, Japan) supplemented with 10% fetal bovine serum, antibiotics, and 10 ng/mL fibroblast growth factor.
Western blot analysis.
Approximately 50 mg of each sample (3 × 106 cells) was homogenized in 0.5 mL RIPA buffer (25 mM Tris, pH 7.4; 50 mM NaCl; 0.5% sodium deoxycholate; 2% NP-40; and 0.2% sodium dodecyl sulfate) with protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 10 μg/mL aprotinin, and 10 μg/mL leupeptin). The homogenate was centrifuged at 14,000 revolutions per minute for 20 minutes at 4°C. The resulting supernatant was collected, and the total protein concentration was determined using the Bradford protein assay (Bio-Rad, Hercules, CA). Western blotting was performed as described previously.25 In brief, 50 μg of protein was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis on 7.5%–12.5% gels. Protein samples then were transferred onto a polyvinylidene difluoride membrane. After blocking in 5% skim milk, the membrane was incubated with the appropriate primary antibody for 1 hour at 4°C using the following concentrations: anti-Ang-1, 1 μg/mL; anti-Ang-2, 1 μg/mL; anti-VEGF, 1 μg/mL; anti-Tie-2, 1 μg/mL; and anti-β-actin, 1:1000 dilution. This step was followed by incubation with the corresponding secondary antibody at a dilution of 1:2000–4000. For detection of the immunocomplex, the Amersham enhanced chemiluminescence detection system (Amersham, Arlington Heights, IL) was used. Protein lysates from human umbilical vein ECs and colon cancer cells served as positive controls for Tie-2 and VEGF, respectively.29–32
Statistical analysis was performed using StatView J-5.0 software (Abacus Concepts, Berkeley, CA). Data are expressed as mean values ± standard deviations. Associations between discrete variables were assessed using Fisher's exact test. Mean values were compared using the Student's t test. The Wilcoxon signed rank test was used to evaluate differences among corresponding objects. Correlation significance was assessed using Pearson's correlation coefficient test. P < .05 was taken to indicate that a given correlation was significant.
Vascularity of CRC Metastases in the Liver.
On computed tomography–arteriography, all metastatic CRCs (n = 14) exhibited ring enhancement, suggesting the presence of an abundant blood supply around the tumors (Fig. 1). Staining of vascular ECs with anti-CD31 antibody was performed to assess the vascular density of the liver tissue and the metastatic CRC. Although positive immunoreactivity for CD31 was noted only in vascular ECs in the Glisson's triangle area (Fig. 2A) and in the central vein in the distant normal liver tissue, there were numerous large blood vessels (diameter [mean ± standard deviation], 51.9 ± 13.6 μm) with positive immunoreactivity in the liver region adjacent to the metastatic CRC (Fig. 2B). Within the metastatic CRC, high vascularity involving short, tortuous blood vessels measuring 28.2 ± 11.6 μm in diameter was observed at the periphery (Fig. 2C), but virtually no CD31-positive ECs were present in the central region (Fig. 2E). The intermediate region between the periphery and the central part of the tumor exhibited intermediate vascularity (Fig. 2D). These tumor vascularity features were observed in all CRC metastases in the liver that were examined. Blood vessel counts were estimated to be 75.5 ± 32.6 per mm2 in the periphery, 33.4 ± 18.2 per mm2 in the intermediate region, and 0 ± 0 per mm2 in the central region (Fig. 2F). There was a significant decrease in vessel count going from the periphery of the tumor to the intermediate region, and also going from the intermediate region to the center of the tumor (P < .001 for each pair of adjacent regions).
ANG2 RNA content in clinical samples.
ANG2 RNA was expressed in 7 of 36 normal mucosae (19.4%), 30 of 36 primary CRCs (83.3%), and 14 of 14 CRC metastases in the liver (100%). ANG2 RNA bands were absent, weak, or strong, whereas β-actin RNA bands were relatively uniform (Fig. 3). High expression of ANG2 RNA (defined by a band density of ANG2 relative to β-actin of > 2) was noted in 3 of 36 normal mucosae (8.3%), 23 of 36 CRCs (63.9%), and 14 of 14 CRC metastases in the liver (100%). Thus, high expression of Ang2 RNA was significantly more common in metastatic CRC than in normal mucosa (P < .0001) or primary CRC (P = .024). Clinicopathologic investigation of primary CRC revealed that larger tumors (P = .023) and the presence of lymph node metastases (P = .033) were significantly associated with high ANG2 RNA expression (Table 3).
Table 3. Association Between Clinicopathologic Features and ANG2 mRNA Expression in Primary Cancer Tissue Samples
Absent or Low
Abbreviations: NS, not significant; mp, muscularis propria; ss, subserosa.
Content of RNA encoding Ang-2 and related molecules in CRC metastases in the liver.
To explore the distribution of cells producing modifiers of angiogenesis (Ang-1, Ang-2, Tie-2, and VEGF), optimal cutting temperature compound–embedded frozen samples of CRC metastases in the liver were dissected via LCM so that RNA could be extracted. Based on CD31 staining results (Fig. 4A), we defined the periphery of the metastatic CRC lesion as the region in which tumor-associated vessels were abundant, and we defined the intermediate region as the area in which tumor-associated vessels were noticeably less abundant. Each region was dissected on eosin-stained serial section (Fig. 4B), but the central region of the tumor was not dissected, because it typically was necrotic. For comparison, the liver region adjacent to the metastatic CRC lesion also was dissected.
ANG2 RNA expression was stronger within the metastatic CRC lesion compared with the bordering liver region. In particular, ANG2 RNA expression in the intermediate region was evident in several cases (Fig. 5; Cases 1–3). Densitometric analysis indicated that there was an increase in ANG2 RNA content going from bordering liver to the tumor periphery, and also going from the periphery to the intermediate region of the tumor (P < .05 for each pair of adjacent regions) (Fig. 6). High expression of ANG1 RNA in the bordering liver region rather than within the metastatic CRC lesion was observed in some cases (Fig. 5; Case 3), but statistical analysis indicated no significant difference between adjacent regions (Fig. 6). High expression of TIE2 RNA was more common in the bordering liver region than within the metastatic CRC lesion (Fig. 5). Statistical analysis indicated that the TIE2 RNA content in bordering liver tissue was significantly greater than the content in the peripheral and intermediate regions of the tumor (Fig. 6; P < .05 in each case). In the analysis of VEGF RNA content, three isoforms, VEGF-121, VEGF-165, and VEGF-189, were considered. When all 3 isoforms were taken into account, there was a significant increase in VEGF RNA content going from bordering liver to the tumor periphery, and also going from the periphery to the intermediate region of the tumor (Fig. 6; P < .05 for each pair of adjacent regions). Considered separately, expression of RNA encoding VEGF-165 and expression of RNA encoding VEGF-121 also increased in similar patterns (data not shown). Expression of RNA encoding VEGF-189 was not quantifiable in some samples.
Because similar expression patterns were observed for VEGF and ANG2 (Fig. 6), we analyzed the correlation between VEGF RNA expression and ANG2 RNA expression. When VEGF expression data were plotted against ANG2 expression data (n = 30; 3 dissected portions from each of 10 samples) (Fig. 7), a significant correlation was noted (r = 0.55; P = .002).
Expression of Ang-2/ANG2 and related molecules/genes at the protein and RNA levels.
Using three colon cancer cell lines and four nonneoplastic colonic mucosa/primary CRC tissue pairs, protein expression of Ang-1, Ang-2, Tie-2, and VEGF was examined by Western blot analysis, and the results of this analysis were compared with the corresponding RNA expression results. With respect to the extent of expression by each of the three cell lines and the relative expression levels in normal mucosa and primary CRC tissue, good agreement was noted for each molecule and corresponding gene at the protein and RNA levels (Fig. 8A,B).
Immunohistochemical analysis of Ang-2 in CRC metastases in the liver.
Immunohistochemical analysis of Ang-2 was performed on 14 paraffin-embedded samples of CRC metastases in the liver. The control section of human placental tissue exhibited positive staining for Ang-2 (Fig. 9A). Liver tissue adjacent to the metastasis generally exhibited weak-to-moderate staining (Fig. 9B), whereas most metastatic CRCs (12 of 14) exhibited moderate-to-strong staining for Ang-2. A gradual increase in expression of the Ang-2 protein going from the periphery to the intermediate region of the metastatic tumor was observed in 11 of 14 cases (Fig. 9C,D). Overall, Ang-2 immunostaining results were well correlated with the corresponding RT-PCR results (data not shown). VEGF staining yielded similar results to those observed for Ang-2 staining (data not shown).
Maturation of blood vessels.
For assessment of the extent of vessel maturation, double-staining for CD31 and α-SMA was performed to detect ECs and PESCs, respectively. Figure 10 shows representative vessels in each tumor region. Mature vessels, such as the portal vein and the hepatic artery, were virtually surrounded by PESCs (Fig. 10A). Within the metastatic CRC lesion, a number of myofibroblastlike cells also expressed α-SMA in the stroma. Vessels at the periphery partially lacked α-SMA expression derived from PESCs (Fig. 10B). Vessels located in the intermediate region were surrounded by PESCs to a lesser extent compared with vessels in the tumor periphery (Fig. 10C). Quantitative analysis indicated that the vessel maturation index was 100% ± 0% in bordering liver, 87.6% ± 20.1% in the tumor periphery, and 64.2% ± 28.1% in the intermediate region of the tumor (Fig. 10D). The difference between each pair of regions was statistically significant (P < .01).
Using computed tomography–angiography and histopathologic methods, we observed abundant tumor vessels at the periphery of the metastatic tumor, where tumor cells generally were viable. In contrast, the center of the metastatic lesion was devoid of CD31-positive tumor vessels and consisted of a considerable amount of necrotic tissue. These findings suggest that tumor angiogenesis at the frontier of the metastatic CRC lesion in the liver may be essential to the consistent growth of malignant cells.
Although it is unknown whether Ang-2 is involved in neovascularization of metastatic CRC, the data from the current study suggest that this may be the case. First, RT-PCR assays indicated that high expression of ANG2 RNA was more common in metastatic CRC tissue than in primary CRC tissue or normal colonic mucosa. Second, LCM analysis revealed that ANG2 RNA was expressed in the peripheral and intermediate regions of metastatic CRC lesions, where tumor angiogenesis was present. In fact, ANG2 RNA levels were higher in the peripheral and intermediate tumor regions than in the bordering liver tissue. Furthermore, some synchronous CRC metastases in the liver exhibited even higher ANG2 RNA expression levels than did the primary CRCs from the corresponding patients (3 of 9 cases [33%]; data not shown).
Clinicopathologic assessment was useful in investigating the putative fundamental action of the ANG2 gene in primary CRC. Our data indicated that, as in thyroid tumors,18 larger primary CRCs had higher ANG2 RNA expression levels. This result is consistent with the finding in recent in vivo studies that transduction of the ANG2 gene into colon or gastric cancer cells produced larger tumors.16, 22 Because these enlarged xenografts exhibited an increased vessel count, it is believed that Ang-2 produced by malignant cells may facilitate neovascularization and contribute to rapid malignant growth. Another notable finding was the positive association between high ANG2 RNA expression and metastasis to the lymph nodes, suggesting a possible role for Ang-2 in metastatic spread.
Of several angiogenic factors (e.g., beta fibroblast growth factor and platelet-derived endothelial cell growth factor), VEGF is the best-characterized proangiogenic agent that is known to be involved in the development of metastatic CRC in the liver.30–32 The usefulness of anti-VEGF agents currently is being tested in clinical trials.33 By using VEGF as a control in the current study, the relevance of Ang-2 in metastatic CRC in the liver was made more evident. The RNA expression pattern of ANG2, but not ANG1, was very similar to that of VEGF (Fig. 6), and we found a significant correlation between expression of ANG2 RNA and expression of VEGF RNA in each tumor region (Fig. 7). Previous studies have demonstrated that Ang-2 promotes angiogenesis synergistically with VEGF in several in vivo models10, 13 and that coexpression of Ang-2 and VEGF often is observed in gastric cancer, glioma, thyroid cancer, and lung cancer.16–18, 21 Together, Ang-2 and VEGF may participate in tumor-associated angiogenesis in liver metastases.
Although RT-PCR assays provide data on gene expression at the RNA level, biologic function is carried out by proteins. Thus, insight into the expression of the angiogenesis-related proteins themselves would be desirable. There is evidence that expression of VEGF messenger RNA (mRNA) is well correlated with VEGF protein expression in human hepatocellular carcinoma.34, 35 It also has been reported that ANG1, ANG2, and TIE2 RNA expression data were concordant with the corresponding protein expression data in human hepatocellular carcinoma and in rheumatoid arthritis.35, 36 We also found good agreement between RNA and protein expression data for the four angiogenesis-related molecules investigated (Fig. 8). Furthermore, immunohistochemical analysis provided confirmatory results regarding Ang-2 and VEGF expression in metastatic CRC. These findings suggest that expression of these proteins may be regulated at a transcriptional level.
The current study raised the question of why RNA expression of ANG2 and VEGF increased going from the tumor periphery to the intermediate region when vascular density was lower in the intermediate region compared with the periphery. To explain this paradox, we posited that insufficient blood supply in the intermediate region might stimulate production of the angiogenic factors in question, given that VEGF is a putative hypoxia-inducible gene.37 The relatively hypoxic environment in the intermediate portion as compared with the periphery was verified by RT-PCR analysis of another hypoxia-inducible gene, GLUT138 (glucose transporter gene-1) (Ogawa M et al., unpublished data, 2003). In support of this hypothesis, there is evidence that hypoxia can induce ANG2 expression in vascular ECs and glioma cells.21, 39–41
Another possible link between Ang-2 and tumor-associated angiogenesis could be inferred from the histopathologic features of the tumor vessels observed. Tumor vessels appeared to be immature, with tortuous morphology and a relatively small luminal size, significantly different from the ordinary straight vessels in normal liver tissue. Other studies also suggested that Ang-2 may be associated with vessel immaturity. The characteristically small luminal size of tumor vessels was reported in ANG2 transgenic mice and in Ang-2-dependent corneal neovascularization in mice.10, 13 It is noteworthy that PESCs were not sufficiently recruited to surround ECs in these Ang-2-associated in vivo models. In addition, it was demonstrated that overexpression of the ANG2 gene produced a lower degree of vessel maturation in in vivo experiments involving gastric cancer cells.16 We consistently found that insufficient recruitment of PESCs around ECs became more evident going from normal liver tissue to the tumor periphery, and also going from the periphery to the intermediate portion of the tumor, and that expression of Ang-2, but not Ang-1, increased accordingly with increasing proximity to the center of the tumor (Figs. 6, 10D). Because Ang-1 maintains and stabilizes mature vessels, these findings suggest that high expression of ANG2 RNA relative to ANG1 RNA may prevent vessel maturation.
In conclusion, we have demonstrated that Ang-2/ANG2 is preferentially expressed at the protein and RNA levels in metastatic CRC in the liver. The current data suggest that Ang-2 may cooperate with VEGF in tumor-associated angiogenesis and thus assist in tumorigenesis of CRC metastases in the liver. Therefore, with respect to anti-VEGF therapy, inhibition of Ang-2 activity may be an alternative or additional strategy in the prevention of CRC-related liver metastasis.
The authors thank Y. Naito and S. Yamane for their work on the preparation of paraffin sections and A. Ueda, M. Yamada, and X. Xu for their experimental assistance.