Functional in vivo optical imaging of tumor angiogenesis, growth, and metastasis prevented by administration of anti‐human VEGF antibody in xenograft model of human fibrosarcoma HT1080 cells

Abstract

Angiogenesis plays a crucial role in cancer progression and metastasis. Thus, blocking tumor angiogenesis is potentially a universal approach to prevent tumor establishment and metastasis. In this study, we used in vivo and ex vivo fluorescence imaging to show that an antihuman vascular endothelial growth factor (VEGF) antibody represses angiogenesis and the growth of primary tumors of human fibrosarcoma HT1080 cells in implanted nude mice. Interestingly, administering the antihuman VEGF antibody reduced the development of new blood vessels and normalized pre‐existing tumor vasculature in HT1080 cell tumors. In addition, antihuman VEGF antibody treatment decreased lung metastasis from the primary tumor, whereas it failed to block lung metastasis in a lung colonization experiment in which tumor cells were injected into the tail vein. These results suggest that VEGF produced by primary HT1080 cell tumors has a crucial effect on lung metastasis. The present study indicates that the in vivo fluorescent microscopy system will be useful to investigate the biology of angiogenesis and test the effectiveness of angiogenesis inhibitors. (Cancer Sci 2009)

Angiogenesis is required for vascular system development in the embryo and normal tissue growth, wound healing, and reproductive function in adults. Angiogenesis also plays a pivotal role in several pathologic disorders, particularly tumorigenesis and metastasis.^{( 1} , ² , ^{3 )} Solid tumors cannot grow beyond 2–3 cubic millimeters without establishing an adequate blood supply because cancer cells only grow within the diffusion boundaries of oxygen and nutrients. Angiogenesis is therefore essential for continued tumor growth.^{( 4 )} In addition, this process promotes progression of the primary lesion by allowing cancer cells to enter the circulation and establish metastatic lesions at distant sites in the body.^{( 5 )} Thus, blocking tumor angiogenesis could provide a universal approach to prevent tumor establishment and metastasis.

Angiogenesis is mediated by many positive and negative regulators, and the degree of angiogenesis depends on the relative balance of these factors. Various cells in the tumor and the microenvironment, including macrophages and fibroblasts, produce angiogenic factors.^{( 5 )} Many molecules have been implicated as positive regulators of angiogenesis, including vascular endothelial growth factor (VEGF), also known as VEGF‐A, acidic fibroblast growth factor (FGF), basic FGF, transforming growth factor (TGF)‐α, TGF‐β, hepatocyte growth factor (HGF, or scatter factor), tumor necrosis factor (TNF)‐α, angiogenin, interleukin (IL)‐8, and the angiopoietins. Among these factors, VEGF is a potent mitogen and survival factor for endothelial cells.^{( 1} , ⁶ , ^{7 )}

Vascular endothelial growth factor (VEGF) is expressed in a number of cancer cell lines^{( 8 )} and in clinical specimens from breast,^{( 9 )} brain,^{( 10 )} ovarian,^{( 11 )} and colon cancers.^{( 12 )} Kim et al. ^{( 13 )} reported that specific VEGF inhibition with an anti‐VEGF antibody decreased tumor neovascularization and substantially inhibited primary tumor growth in vivo, confirming that VEGF is an important mediator of tumor angiogenesis. Previous studies also showed that VEGF inhibition alone is sufficient to prevent tumor dissemination.^{( 14} , ^{15 )} Subsequently, many tumor cell lines were found to be inhibited in vivo by various anti‐VEGF treatments, including small‐molecule inhibitors of VEGFR signaling, antisense oligonucleotides, and antibodies to VEGFR‐2.^{( 7} , ^{16 )}

The continual development of more efficient angiogenesis inhibitors has increased the need to develop and validate techniques that measure physiologic and molecular surrogate markers of angiogenesis. However, there are few techniques that measure angiogenesis in vivo. In the present study, we have developed a method to measure both tumor angiogenesis and lung metastasis within the same animal using green fluorescent protein (GFP)–expressing tumor cells and near‐infrared (NIR) probes, and investigated the effect of antihuman VEGF antibody treatment on the angiogenesis, growth, and lung metastasis of human fibrosarcoma HT1080 cells in mice.

Materials and Methods

Cell lines.  HT1080, a human fibrosarcoma cell line, and DU4475, a human breast cancer cell line, were obtained from the American Tissue Culture Collection (ATCC; Manassas, VA, USA). HT1080 cells and DU4475 cells were maintained according to the experimental procedure of ATCC.

Lentiviral production and infection.  To establish HT1080 cells that express GFP, we used a lentiviral expression system. Viral production and infection were performed as described previously.^{( 17 )}

Quantitative real‐time reverse transcription–polymerase chain reaction (RT‐PCR).  Quantitative real‐time RT‐PCR was performed as described.^{( 18 )} The following VEGF primer sequences were used: forward, 5′‐AGTGGTCCCAGGCTGCAC‐3′; reverse, 5′‐TCCATGAACTTCACCACTTCGT‐3′. VEGF primers were designed in exons 1 and 2 so that all of VEGF isoforms could be measured.^{( 19 )} All samples were run in duplicate in each experiment.

Cell proliferation assay.  HT1080 cells were seeded in duplicate at a density of 1 × 10⁴ cells/well in six‐well plates. After starvation, cells were incubated with or without 2 μg/mL of antihuman VEGF antibody (R&D Systems, Minneapolis, MN, USA) for 72 h, followed by trypsinization and counting.

Animal models.  All animal procedures were performed in the animal experiment room of the Japanese Foundation for Cancer Research (JFCR) according to the guidelines proposed by the Science Council of Japan. Female BALB/c nu/nu mice (8 weeks) were purchased from Charles River Japan (Kanagawa, Japan). In the angiogenesis and spontaneous metastasis experiment, 2 × 10⁶ HT1080 cells or DU4475 cells were inoculated into the mammary glands of mice. In the lung colonization experiment, 1 × 10⁶ HT1080 cells were inoculated into the lateral tail vein of mice. All in vivo experiments were performed on anesthetized (gas anesthesia [isoflurane 1.5–2%]) mice.

In vivo imaging and quantification of tumor vasculature.  In vivo imaging was performed with two in vivo imaging systems. The OV100 (Olympus, Tokyo, Japan) is a variable‐magnification whole‐mouse imaging system^{( 20 )} equipped with a band pass filter (610–645 nm) and a 150‐W Xenon light source. Images were acquired with a × 0.27 objective lens. All data were taken with a monochrome digital charge‐coupled device (CCD) camera (F‐View II; Olympus Soft Imaging Solutions, Munster, Germany). The IV100 (Olympus) is an intravital laser scanning microscope with a Helium‐Neon (633 nm) laser source.^{( 21} , ^{22 )} Images were acquired using a × 4 objective lens (NA 0.13). The tumor vasculature was imaged by injecting 30 μL of AngioSense‐IVM 680 (AS‐IVM 680; VisEn Medical, Woburn, MA, USA) intravenously. AngioSense‐IVM680 is kept in the mouse body for about 1 h. For quantitative analysis, single images of tumor vasculature obtained by the IV100, and the vasculature area was determined using Image J software (Research Services Branch, National Institute of Mental Health, Bethesda, MD, USA).

Measurement of microvessel density (MVD).  Tumors were excised, snap frozen, and sectioned (10‐μm thickness). After fixation and endogenous peroxidase blocking, sections were incubated with mouse CD31 monoclonal antibody (PharMingen, San Diego, CA, USA) overnight at 4°C, followed by the appropriate Histofine Simple Stain Mouse MAX‐PO (Nichirei, Tokyo, Japan) secondary antibody. The sections were counterstained with hematoxylin. To quantify MVD, the sections were scanned at low power (×100) to identify the most vascularized areas. Microvessels were counted at high power fields (×400) in the three most vascularized areas by two blinded investigators, and the two scores were averaged. A single microvessel was defined as a discrete cluster or single cell, positive for CD31, and the presence of a lumen was required for scoring as a microvessel.^{( 23 )}

In vivo antihuman VEGF neutralizing antibody treatment.  The antihuman VEGF antibody which reacts only with human VEGF was diluted in PBS to a 200‐μL volume and was intraperitoneally injected into mice twice weekly at a 3 mg/kg dose. As a control, PBS was used. Twenty‐three tumor‐bearing animals were randomly assigned to a treatment group (control, n = 10; antihuman VEGF antibody, n = 13). The size of the primary tumor was measured using calipers, and tumor volume was calculated using the following formula: tumor volume = ab ²/2, in which a is the longest diameter and b is the shortest diameter of the tumor. All animals were sacrificed 24 days post inoculation (i.e. after 3 weeks of antibody treatment) and examined for spontaneous lung metastasis. Twenty tumor‐bearing animals were randomly assigned to a treatment group (control, n = 10; antihuman VEGF antibody, n = 10). All animals were sacrificed 13 days post inoculation (i.e. after 2 weeks of antibody treatment) and examined for lung colonization.

Ex vivo imaging and quantification of lung metastases.  Mice treated with PBS (spontaneous experiment, n = 9; lung colonization experiment, n = 10) or antihuman VEGF antibody (spontaneous experiment, n = 13; lung colonization experiment, n = 10) were sacrificed and their lungs were examined by the OV100 ex vivo. A GFP band pass filter (460–480 nm) was used to acquire GFP‐positive foci. In the spontaneous experiment, images were acquired with a × 0.56 objective lens. The number of GFP‐positive micrometastases was counted by two investigators in a blinded fashion. In the lung colonization experiments, images were acquired with a × 0.27 objective lens. Images were processed into binary images. The area of GFP‐positive micrometastasis was determined using Image J software.

Statistical analysis.  To determine a linear association, we calculated the Pearson’s correlation. In order to show a distribution of the data, we used a boxplot^{( 24 )} which consists of the smallest observation, lower quartile (Q1), median, upper quartile (Q3), and largest observation. The Q1 is the lower side and Q3 is the upper side of the rectangle in the boxplot. The bar in the rectangle indicates the median of the data. Welch’s t‐test was performed to compare two distributions. To compare tumor growth curves, we fit an autoregressive random error model with a linear mean assuming a different slope but a common intercept for each antibody (+/−) type.^{( 25 )} The autoregressive random error model was implemented and tested using the statistical software R 2.4.1 with agce package (R Development Core Team, Vienna, Austria). The hypothesis that the growth curves would have equal slopes was tested by the likelihood ratio test with a significance level of 0.05.

Results

Evaluation of angiogenic activity of distinct cancer cell lines by in vivo fluorescent imaging.  Real‐time in vivo fluorescent imaging at a cellular and subcellular resolution is a developing technique that will be essential to better understand complex biological processes. To quantitatively characterize tumor vascularization and monitor the response of the human fibrosarcoma cell line HT1080 to anti‐angiogenic therapy, we first evaluated the potential of in vivo fluorescent imaging systems (OV100 and IV100) in combination with the NIR fluorescent probe AS‐IVM 680. As a control, we used the human mammary adenocarcinoma cell line DU4475, which has reduced VEGF mRNA levels compared to HT1080 (Fig. 1A) in vitro. Nude mice were implanted with DU4475 cells and HT1080 cells, and examined 15 days post inoculation for tumor vasculature using the in vivo imaging system described in the ‘Materials and Methods’. Angiogenesis was highly abundant in the HT1080 cell tumor but moderate in the DU4475 cell tumor (Fig. 1B). Blood vessels in the HT1080 cell tumor were tortuous and dilated. After in vivo imaging, we performed a conventional pathological examination on the same tumors. There was a higher density of microvessel clusters in the HT1080 cell tumor than in the DU4475 cell tumor (Fig. 1C), confirming the differences seen by vascular imaging.

Figure 1.

 Comparison of the angiogenic activity of distinct cancer cell lines by in vivo fluorescent imaging. (A) Quantitative real‐time RT‐PCR analysis of vascular endothelial growth factor (VEGF) mRNA levels in HT1080 cells and DU4475 cells. Each value is normalized to GAPDH expression. The increased VEGF mRNA levels in HT1080 cells compared to DU4475 cells are evident. Error bars represent SD. (B) In vivo imaging of tumor vasculature. Tumor vasculature was imaged 15 days after inoculation of DU4475 cells (left panels) or HT1080 cells (right panels) using AS‐IVM 680. Top panels show vasculature around the tumor imaged by the OV100 (arrowheads). Asterisks indicate AS‐IVM 680 diffusion in the tumor. Scale bar, 5 mm. Bottom panels show vasculature in the tumor imaged by the IV100. Scale bar, 500 μm. (C) The same tumors as in (B) after immunohistochemical staining for CD31. Left and right panels show DU4475 and HT1080 cell tumors, respectively. Scale bar, 30 μm.

Time‐course analysis of the angiogenic activity of HT1080 cells by in vivo fluorescent imaging.  We next monitored angiogenesis in HT1080 cells using serial imaging. On day 1 post inoculation, some vasculatures were observed in the HT1080 cell tumors, but not in the surrounding tissues (Fig. 2A, left two panels). On day 8 post inoculation, vigorous vascularization was observed in the tumor and surrounding tissues (Fig. 2A, middle two panels). On day 14 post inoculation, the HT1080 cell tumor contained numerous, irregular vessels that were tortuous and dilated, and vigorous neovascularization was also observed in the surrounding tissues (Fig. 2A, right two panels). To confirm the differing degrees of vascularization on each progressive day of angiogenesis, a conventional pathological examination of CD31 staining (MVD) was performed on the same sections after in vivo imaging on each day (Fig. 2B). A statistically significant correlation between in vivo imaging (vasculature area) and MVD was observed with a Pearson’s correlation coefficient of 0.867 (P = 0.0021) (Fig. 2C). These results suggest that the NIR laser scanning microscope system in combination with NIR probes is a powerful method for real‐time continuous imaging of tumor angiogenesis.

Figure 2.

 Serial in vivo imaging and quantification of tumor vasculature in HT1080 cell tumors. (A) Serial images during the progression of angiogenesis. In vivo imaging of tumor vasculature was repeatedly performed at 1 day (left panels), 8 days (middle panels), and 14 days (right panels) post inoculation using AS‐IVM 680. Top panels show vasculature around the tumor imaged by the OV100 (arrowheads). Asterisks indicate AS‐IVM 680 diffusion in the tumor. Scale bar, 5 mm. Bottom panels show vasculature in the tumor imaged by the IV100. Scale bar, 500 μm. (B) Complementary in vivo imaging and immunohistochemical analysis at different time points. The same tumors were examined by the IV100 (top panels), and subsequently, immunostained for CD31 (bottom panels) at 10 days (left panels), 12 days (middle panels), and 15 days (right panels) after inoculation. White bar, 500 μm. Black bar, 30 μm. (C) Validation. The graph shows the correlation between the IV100 measurements of tumor vasculature and microvessel density (MVD) measurements (n = 9 different tumors). Blue, yellow, and red points respectively indicate measurements at 10 days, 12 days, and 15 days after inoculation. The Pearson’s correlation coefficient was 0.867 (P = 0.0021). Quantification of vasculature area and MVD was performed as described in the ‘Materials and Methods’.

Effect of systemic antihuman VEGF antibody treatment on HT1080 cell tumor angiogenesis and growth.  VEGF is an important mediator of tumor angiogenesis, and specific inhibition of VEGF decreases vascularization of various cancer cells in vivo. Therefore, we examined the therapeutic effects of an antihuman VEGF antibody on HT1080 cell angiogenesis by combining NIR fluorescent probes with the molecular imaging systems described in the ‘Materials and Methods’. As shown in Figure 3(A), antihuman VEGF antibody treatment reduced outgrowth of pre‐existing tumor vasculature and new blood vessel development of HT1080 cell tumors detected by the OV100 and IV100. Normal vessels in the ear of the antibody‐treated mice were not affected (data not shown). The median vascular areas for antihuman VEGF antibody‐treated mice and control mice were 0.00556 cm² and 0.00193 cm², respectively with a statistically significant (P = 0.000196) difference between the two groups as determined by Welch’s t‐test (Fig. 3B). These results suggest that the imaging system is useful to evaluate anti‐angiogenic therapies. In addition, as shown in Figure 3(C), the same vasculatures could be observed serially using this system.

Figure 3.

 Effect of antihuman vascular endothelial growth factor (VEGF) antibody treatment on tumor angiogenesis and growth. (A) Serial images of tumor vasculature during treatment of antihuman VEGF antibody. Tumor vasculature of control mice (left panels) or antibody‐treated mice (right panels) was imaged using AS‐IVM 680. These images were acquired 3 days post inoculation, that is before therapy initiation, (top panels) and 10 days post inoculation, that is after 1 week of therapy, (bottom panels). Vasculature around the tumor was imaged by the OV100 (arrowheads). Asterisks indicate AS‐IVM 680 diffusion in the tumor. Scale bar, 5 mm. Vasculature in the tumor was imaged by the IV100. Scale bar, 500 μm. (B) Quantification of tumor angiogenesis. Values show the tumor vasculature area at 10 days post inoculation minus that at 3 days post inoculation. There was a statistically significant decrease in tumor angiogenesis in mice treated with antihuman VEGF antibody compared to control mice (P = 0.000196). The boxplots show the smallest observation, lower quartile, median (the bar in the box), upper quartile, and largest observation. (C) Serial images of the same vasculatures at the different time points during treatment of the antibody. Vasculature of untreated tumor (left panels) or tumor (middle panels) and ear (right panel) of treated mice was imaged by the IV100 using AS‐IVM 680. These images were acquired at 3 days post inoculation, that is before therapy initiation, (top panels) and 11–12 days post inoculation, that is after 1 week of therapy, (bottom panels). The boxed area indicates the same vasculatures at the different time points. Scale bar, 500 μm. (D) Mean tumor growth curves for antihuman VEGF antibody‐treated mice and control mice. Systemic treatment with antihuman VEGF antibody significantly inhibited tumor growth compared to untreated mice (P < 0.0001). (E) The effects of antihuman VEGF antibody on the cell growth of HT1080 cells. Cultured HT1080 cells were incubated with (+) or without (−) antihuman VEGF antibody for 72 h, and cell numbers were counted. Error bars represent SD.

We next examined tumor growth in a xenograft model of HT1080 cell lines. As shown in Figure 3(D), HT1080 tumors grew more slowly in the antihuman VEGF antibody‐treated group compared to the control group. The initial mean tumor volume in the control group was 13.8 mm³ and increased to 1810.4 mm³ after 3 weeks, while tumor volumes in the treated group increased from 12.5 mm³ to 437.5 mm³. We fit an autoregressive random error model with a linear mean, assuming a different slope but a common intercept for each antibody (+/−) type. The null hypothesis that the growth curves would have equal slopes was rejected with a P‐value less than 0.0001, and there was strong evidence of reduced tumor growth according to antibody treatment. Notably, antihuman VEGF antibody did not affect HT1080 cell growth in vitro (Fig. 3E). These results suggested that antihuman VEGF antibody treatment reduces HT1080 cell growth in vivo by reducing tumor angiogenesis.

Effect of systemic antihuman VEGF antibody treatment on spontaneous lung metastasis of HT1080 cell tumors.  The formation of new blood vessels from the pre‐existing vasculature is thought to be necessary not only for primary tumor growth but also for metastasis. To clarify the role of angiogenesis in cancer metastasis, we examined the effect of antihuman VEGF antibody treatment on spontaneous lung metastasis of HT1080 cells. In the lungs of HT1080 tumor‐bearing mice, fluorescent nodules were detected by the OV100 (Fig. 4A, left panel). These nodules had a minimal size of ∼1 mm and could not be detected by conventional macroscopy. They were histologically confirmed to be cancer cells by H&E staining (Fig. 4A, right panel). We then examined the effect of antihuman VEGF antibody treatment on lung micrometastasis using the ex vivo imaging system. Representative data shown in Figure 4(B) demonstrate that antihuman VEGF antibody treatment inhibited spontaneous lung metastasis of the HT1080 primary tumor. In Figure 4(C), the boxplots of numbers of lung metastasis for antibody (−) and (+) are shown. The average number of lung metastasis in the treatment and control groups was 11.67 and 2.08, respectively with a significant difference between the two groups (P = 0.0043 by Welch’s t‐test).

Figure 4.

 Effect of antihuman vascular endothelial growth factor (VEGF) antibody treatment on spontaneous lung metastasis. (A) Complementary ex vivo imaging and immunohistochemical analysis. Imaging and analysis were performed 24 days after inoculation of GFP‐labeled HT1080 cells. Left panel shows fluorescent tumor micrometastases detected by the OV100. Asterisk indicates tissue autofluorescence. Scale bar, 1 mm. Right panel shows H&E staining of the boxed area in the left panel with arrows indicating metastatic foci. Scale bar, 60 μm. (B) Ex vivo images of fluorescent lung micrometastases in control mice (left panel) or antihuman VEGF antibody‐treated mice (right panel). Ex vivo imaging was performed 24 days after inoculation (i.e. after 3 weeks of treatment) by the OV100. Arrowheads indicate metastatic foci. Asterisks indicate tissue autofluorescence. Scale bar, 1 mm. (C) Quantification of spontaneous lung micrometastasis. The graph shows the spontaneous lung micrometastatic efficacy in antibody‐treated mice. The boxplots show the smallest observation, lower quartile, median (the bar in the box), upper quartile, and largest observation.

Effect of systemic antihuman VEGF antibody treatment on experimental lung metastasis of HT1080 cell tumors.  It is unclear whether antihuman VEGF antibody inhibits metastases by affecting the primary site and/or the metastatic site. We thus examined the effect of antihuman VEGF antibody treatment on another model of lung metastasis in which HT1080 tumor cells were injected into the tail vein. In the lungs of HT1080 tumor‐bearing mice, fluorescent nodules were detected by the OV100 (Fig. 5A). These nodules had a maximum size of ∼1 mm. In Figure 5(B), boxplots of the area of GFP‐positive foci for antibody (−) and (+) are shown. The average area occupied with GFP‐positive foci in the treatment group and control group was 14.5% and 8.7%, respectively.

Figure 5.

 Effect of antihuman vascular endothelial growth factor (VEGF) antibody treatment on experimental lung metastasis. (A) Ex vivo images of fluorescent lung metastases of control mice (left panel) or antihuman VEGF antibody‐treated mice (right panel). Ex vivo imaging was performed 13 days after inoculation of GFP‐labeled HT1080 cells (i.e. after 2 week of treatment) by the OV100. Scale bar, 1 mm. (B) Quantification of experimental lung metastasis. The plots show the percentage of GFP‐positive foci within an area in control mice and antihuman VEGF antibody‐treated mice. Quantification of GFP‐positive foci within an occupied area was performed. The boxplots show the smallest observation, lower quartile, median (the bar in the box), upper quartile, and largest observation.

Association among tumor angiogenesis, tumor growth, and lung metastasis.  We showed an association between tumor angiogenesis, tumor growth, and lung metastasis. For this analysis, we applied a logarithmic transformation to the number of lung metastasis to stabilize its variance. The correlation between IV100 measurements of tumor blood area and tumor volume was 0.834 (P < 0.00001) (Fig. 6A). The correlation between IV100 measurements of tumor blood area and the number of pulmonary metastatic foci was 0.824 (P = 0.00198) (Fig. 6B). The correlation between tumor volume and the number of lung metastasis was 0.759 (P < 0.0001) (Fig. 6C).

Figure 6.

 Correlation between tumor angiogenesis, tumor growth and spontaneous lung metastasis. (A) The correlation between the IV100 measurements of tumor vasculature area and tumor volume (P < 0.00001). (B) The correlation between the IV100 measurements of tumor vasculature area and the number of spontaneous lung metastasis (P = 0.00198). (C) The correlation between the tumor volume and the number of spontaneous lung metastasis (P < 0.0001).

Discussion

There is growing interest in applying optical imaging approaches to study disease processes and complex biology in vivo. In particular, fluorescent imaging has been recently developed in the molecular and cellular biology field. To investigate the effect of antihuman VEGF antibody on tumor angiogenesis, growth, and lung micrometastasis within the same animal, we used fluorescence imaging technique that combines in vivo near‐infrared fluorescence imaging and ex vivo GFP fluorescence imaging. Although the conventional histological measurement of MVD is widely used to validate angiogenesis measured in vivo,^{( 26 )} this method is primarily applied ex vivo and prevents the serial study of real‐time angiogenesis in vivo. In contrast, the in vivo fluorescent imaging system is noninvasive and allows longitudinal studies in a single animal, which can reduce the number of experimental animals without compromising statistical significance. In addition, high‐resolution in vivo fluorescent imaging provides detailed information such as vessel structure (Fig. 2A). Moreover, the correlation between in vivo fluorescent imaging and MVD was statistically significant (Fig. 2C). Thus, this system could be used to analyze both qualitative and quantitative changes simultaneously within the same animal.

Interestingly, although antihuman VEGF antibody treatment reduced the development of new blood vessel and the outgrowth of pre‐existing tumor vasculature in HT1080 cell tumors, it did not decrease numbers of established tumor vasculature (Fig. 3C). In addition, the morphology of established blood vessels in the tumor changed, and the vascular density and vessel diameter decreased (data not shown). Tong et al. ^{( 27 )} showed that blocking VEGF signaling by DC101 (a VEGF‐receptor‐2 antibody) decreases interstitial fluid pressure, not by restoring lymphatic function, but by producing a morphologically and functionally ‘normalized’ vascular network. Moreover, they demonstrated that the normalization process prunes immature vessels and improves the integrity and function of the remaining vasculature by enhancing the perivascular cell and basement membrane coverage. Our results suggest that the antihuman VEGF antibody also normalizes the tumor vascular network of HT1080 cell tumors. Blocking VEGF signaling may target sensitized endothelial cells, collectively destabilizing pre‐existing tumor vasculature and inhibiting on‐going angiogenesis. Of note, recent clinical practice revealed that therapy with angiogenesis inhibitors often does not prolong survival of cancer patients for more than months, because tumors elicit evasive resistance.^{( 28 )} Pàez‐Ribes et al. and Ebos et al. ^{( 29} , ^{30 )} reported that VEGF inhibitors reduce primary tumor growth but promote tumor invasiveness and metastasis. Future studies should focus on how VEGF inhibitors can induce divergent effects on primary tumor and metastasis.

In addition to blocking angiogenesis, there was a significant reduction in tumor growth with antihuman VEGF antibody treatment (Fig. 3D). Previous studies have suggested that inhibiting tumor‐secreted VEGF limits primary tumor growth of sarcoma cell lines by inhibiting host angiogenesis.^{( 13 )} Our results suggest a possible mechanism for inhibition of HT1080 primary tumor growth. Since antihuman VEGF antibodies had no direct effect on tumor cell proliferation in vitro (data not shown), human fibrosarcomas might require the host VEGF‐dependent neovascular response for primary tumor growth.

In this study, we used GFP‐expressing HT1080 cells to study lung micrometastasis using ex vivo fluorescent imaging. Conventional micrometastasis measurements are cumbersome pathological examinations, such as histology and immunohistochemistry or intricate molecular techniques, for example RT‐PCR. In contrast to these methods, ex vivo fluorescent imaging is rapid and inexpensive as no substrate is required, and provides more information including the size and number of metastatic foci. This technique has such high resolution and sensitivity that metastatic foci smaller than 1 mm (Fig. 4A) can be detected. Thus, ex vivo GFP visualization of tumor micrometastasis is much more facile than the conventional examination.

Increased vascularity is thought to increase both tumor growth and metastatic opportunities.^{( 31 )} It has been reported that the overexpression of VEGF correlates with high microvascular density and the frequency of advanced stage pancreatic cancer.^{( 32 )} In the present study, we showed that administration of antihuman VEGF antibody reduced the spontaneous lung micrometastasis of HT1080 cells, suggesting that VEGF plays a crucial role in HT1080 metastasis. The size of lung metastasis detected by the system is too small to be affected by angiogenesis in the metastasis sites. Thus, the effect of the antibody on metastasis is likely due to inhibition of primary tumor microvessels or tumor cell entry into the circulation, but not inhibition of tumor growth at metastatic sites. As such, antihuman VEGF antibody failed to block metastasis in the lung colonization experiment in which tumor cells were injected into the tail vein (Fig. 5). In addition, it is possible that primary HT1080 cell tumors have a crucial effect on lung tissues via VEGF signaling during lung metastasis. Of note, Hiratsuka et al. ^{( 33 )} proposed a crucial role for the VEGF‐receptor/matrix metalloproteinase (MMP) system in tumor metastasis. They demonstrated that the primary tumor themselves determine the specific site of metastasis via VEGF signaling by selectively inducing MMP9 expression in lung endothelial cells and macrophages, thereby promoting the invasion of tumor cells, preferentially lung. Future studies should focus on target cells of VEGF during the lung metastasis of HT1080 cells.

In conclusion, in vivo fluorescent imaging, in conjunction with appropriate fluorescent probes, allows quantitative and qualitative visualization of tumor angiogenesis and lung metastasis. This method will be useful for investigating angiogenesis biology and testing the effectiveness of angiogenesis inhibitors.