Tenascin-C regulates angiogenesis in tumor through the regulation of vascular endothelial growth factor expression

Authors

  • Keiichiro Tanaka,

    1. Department of Surgery, Jikei University School of Medicine, Tokyo, Japan
    2. Department of Anatomy, Jikei University School of Medicine, Tokyo, Japan
    3. Experimental Animal Division, Bio Resource Center, RIKEN, Ibaraki, Japan
    Search for more papers by this author
  • Noriko Hiraiwa,

    1. Experimental Animal Division, Bio Resource Center, RIKEN, Ibaraki, Japan
    Search for more papers by this author
  • Hisashi Hashimoto,

    1. Department of Anatomy, Jikei University School of Medicine, Tokyo, Japan
    2. Experimental Animal Division, Bio Resource Center, RIKEN, Ibaraki, Japan
    Search for more papers by this author
  • Yoji Yamazaki,

    1. Department of Surgery, Jikei University School of Medicine, Tokyo, Japan
    Search for more papers by this author
  • Moriaki Kusakabe

    Corresponding author
    1. Department of Anatomy, Jikei University School of Medicine, Tokyo, Japan
    2. ANB Tsukuba Institute, Aloka Co. Ltd., Ibaraki, Japan
    • ANB Tsukuba Laboratory, Aloka Co. Ltd., 1103 Fukaya, Kasumigaura-Machi, Niihari-Gun, Ibaraki, 300-0134, Japan
    Search for more papers by this author
    • Fax: +81-29-834-6177


Abstract

In order to verify whether tenascin-C (TN-C) is involved in angiogenesis as an extracellular signal molecule during tumorigenesis, cancerous cell transplantation experiments and coculture experiments were carried out, focusing on the regulation of vascular endothelial growth factor (VEGF). The A375 human melanoma cells introduced the GFP gene (A375-GFP), implanted subcutaneously into BALB/cA nude (WT) and TN-C knockout BALB/cA nude (TNKO) congenic mice. Furthermore, coculture experiments between A375-GFP and embryonic mesenchyme, which was prepared from both genotypes, were carried out to investigate the molecular mechanism in the cell-cell interactions. Both the content of TN-C and that of VEGF in the tumor and the conditioned medium were analyzed by the sandwich ELISA method. Seven days after transplantation of the A375-GFP, capillary nets became far more abundant in the tumors grown in WT mice than those in TNKO mice. Interestingly, VEGF and TN-C expressions showed antithetical expression patterns between the tumors in WT mice and those in TNKO mice. This peculiar phenomenon seems to be caused by a time lag prior to the onset of the mesenchymal regulation for the TN-C expression of A375-GFP. The coculture experiments revealed that WT mesenchyme had a much stronger effect than TNKO mesenchyme on both TN-C and VEGF expression. However, the defects of TNKO mesenchyme were restored in all cases by additional TN-C. These results clearly indicated that the expressions of both TN-C and VEGF depend on the surrounding mesenchyme, and that the function of mesenchyme is regulated by its own mesenchymal TN-C. In conclusion, the present data suggest that the matrix microenvironment organized by the host mesenchyme is very important for angiogenesis in tumor development. © 2003 Wiley-Liss, Inc.

It is well known that extracellular matrices (ECMs) produced by both epithelial cells and surrounding mesenchymal cells are involved in cell proliferation, cancer invasion, metastasis and secondary tumor formation. Tenascin-C (TN-C)1 is an extracellular matrix glycoprotein consisting of 6 disulfide-linked subunits and is called an oncofetal molecule because of its unique expression. It is well accepted that TN-C is heavily involved in cell-cell interactions such as the stimulation of cancerous cell mitogenesis, cell adhesion and antiadhesion.2, 3 TN-C especially inhibits the cell adhesion mediated by fibronectin and may play an important role in the regulation of dynamic changes in ECM.4 In 1992, TN-C gene knockout mice (TNKO) were produced by homologous recombination. Unfortunately, they developed normally without any major phenotypic defects, leading us to speculate that TN-C may play an insignificant role in normal development.5 Even though TN-C expression is very limited in adult normal tissues, it is strongly reexpressed in the stroma of malignant tumors, inflamed tissues and healing wounds. Therefore, we have focused on the function of TN-C in vivo during instances of tissue remodeling such as carcinogenesis and wound healing.6, 7, 8

It has been reported that stromal TN-C regulates the stimulation of epithelial-mesenchymal interactions.9, 10, 11 Previous reports12, 13, 14 prompted us to hypothesize that TN-C may play crucial roles during tissue remodeling in such processes as embryogenesis, carcinogenesis and wound healing. Many investigators interested in its expression profiles expected it to develop into a prognostic factor.15 Furthermore, recent reports have clearly demonstrated that the TN-C knockout mouse exhibited extraordinary behavior during the tissue remodeling process in cancer growth and wound healing.16, 17, 18, 19, 20, 21, 22 Angiogenesis, the growth of new capillaries from preexisting blood vessels, is essential for cancers to grow beyond minimal size.23 Folkman24 proposed that tumor growth and metastasis are angiogenesis-dependent; hence, blocking angiogenesis could be a significant strategy for arresting tumor growth. There is considerable evidence that vascular endothelial growth factor (VEGF) is a major tumor angiogenic factor, although angiogenesis is directly mediated by a variety of factors. High levels of VEGF are generally associated with hypoxia, an excess of soluble-inducing factors or unregulated VEGF expression.25, 26, 27, 28, 29 Vascular endothelial cells in the tumor and peritumor tissues also appear to upregulate the expression of VEGF receptors, while many cytokines and growth factors upregulate VEGF mRNA or induce VEGF release. Since tumor growth heavily depends on the vascularization ability, it is worthwhile analyzing the molecular mechanisms of angiogenesis in tumors.

In this study, in order to verify the function of stromal TN-C in the angiogenesis occurring during the early stage of tumor growth, we have performed both cancerous cell transplant experiments and coculture experiments. We began by generating TNKO BALB/cA congenic nude mouse strains, which have a higher percentage of BALB/cA in their genetic background. The GFP gene was then introduced into the A375 human melanoma cell line so as to visualize implanted cancerous cells in the body. Next, the newly formed capillary net of the tumor was analyzed 3-dimensionally by means of a cardiac perfusion of rhodamine-labeled gelatin.30 Furthermore, to verify whether or not the expressions of both human TN-C and human VEGF depend on the mesenchyme, a coculture system using the embryonic mesenchymes prepared from both genotypes of mouse embryos was employed. The expressions of both human TN-C and human VEGF were analyzed by the sandwich ELISA method. The present results indicated that a lack of TN-C from the host mesenchyme caused a dysfunction of the mesenchyme, affecting the expressions of both TN-C and VEGF by the tumor. Furthermore, the data suggested that stromal TN-C plays an important role in the regulation of VEGF expression by cancerous cells.

MATERIAL AND METHODS

Animals

A TN-C gene knockout BALB/cA congenic strain (BALB/cA/TN−/−) was newly established from the TN-C gene knockout mouse by backcrossing more than 15 times. This strain was crossed with the BALB/cA nude mouse (Clea, Tokyo, Japan) and bred over at least 8 generations to obtain a TN-C gene knockout BALB/cA nude congenic strain (BALB/cA-nu/TN−/−) with stable genotyping. The BALB/cA-nu/TN+/+ (WT) and BALB/cA-nu/TN−/− (TNKO) strains were maintained in the Experimental Animal Division of the Bio Resource Center at the RIKEN Tsukuba Institute. Eight-week-old male mice from each strain were used in this study, and all animal experiments were carried out in accordance with the guidelines of the RIKEN Institute's Animal Experimentation Committee.

Cells and introduction of GFP gene

The A375 human melanoma cell line was obtained from the American Tissue Type Culture Collection (CRL-1619; Rockville, MD) and maintained in Iscove's modified Dulbecco's medium (IMDM: Sigma, St. Louis, MD) supplemented with 10% fetal bovine serum (Filtron, Brooklyn, Australia) at 37°C in a humidified atmosphere of 5% CO2 and 95% air. The enhanced GFP gene constructed with cytomegalovirus (CMV) promoter containing the neomycin-resistant gene (pEGFP-N1 Vector) was purchased from Clontech (Palo Alto, CA). Gene introduction into A375 cells was performed by means of the lipofectamine method using the lipofectamine reagent protocol (Gibco-BRL, Rockville, MD). After introduction of the GFP gene, the cells were subcultured for 2 weeks for selective screening in IMDM containing neomycin (1,000 μg/ml; Gibco-BRL). A single cell emitting the most intense fluorescence was transferred into each well of a 96-well plate. It was then cultured in IMDM containing 10% FCS. The established A375-GFP cells showed the biological characteristics such as cell shape and growth rate identical to those of the original A375 cell line and exhibited an intense fluorescence of GFP after 4 weeks in vitro.

Transplant experiments

Healthy 90% confluent A375-GFP cells were trypsinized with 0.05% trypsin containing 0.53 mM EDTA-4Na (Gibco-BRL). After the inactivation of triptic activity by adding the medium with 10% FCS, they were centrifuged and resuspended into the fresh medium. Cell numbers were then counted using a hemocytometer. Two million cells were resuspended into 200 μl of physiologic saline and then subcutaneously injected with a 26-gauge needle under ether anesthesia. On days 1, 3, 5, 7, 14 and 21 after inoculation, 5 animals from each group were sacrificed in an experiment. Each tumor was weighed and subjected to the series of experiments described below.

Antibodies

Two rat monoclonal antihuman TN-C antibodies, 7-13 and 36-13-6, developed in our laboratory were used for the sandwich ELISA method. The 7-13 and 36-13-6 can recognize fibronectin type III repeats (3–5) and alternative splicing sites (A–D), respectively. The 7-13 is human-specific, whereas the 36-13-6 recognizes both human and mouse TN-C. The former was used as a first antibody to trap only human TN-C, the latter as a secondary antibody after biotinylation. Rabbit antihuman VEGF antibody (Sc-507; Santa Cruz, Santa Cruz, CA) and biotinylated rabbit antihuman VEGF antibody (Sc-152; Santa Cruz) were purchased. The Sc-507 recognizes both human and mouse VEGF, whereas the Sc-152 is human-specific.

Preparation of rhodamine-labeled gelatin solution

The rhodamine-labeled gelatin was prepared according to previously published method.30, 31 Briefly, 50 mg of rhodamine B isothiocyanate (RITC; Sigma) dissolved in 1 ml of dimethylsulfoxide (DMSO; Sigma) was added to 40 ml of the 20% aqueous gelatin (G-9382; Sigma) solution (pH 9.0). This mixture was shaded by an aluminum foil and gently stirred overnight at 37°C. The gelatin mixture was poured into 500 ml of 100% ethyl alcohol to remove the free rhodamine. The gelatin fibers were then reconstructed by dehydration and rinsed with 100% ethyl alcohol several times until the ethyl alcohol became colorless. They were then dried and dissolved in PBS to make a 10% gelatin solution. An aliquot of the solution was kept in the dark at 4°C until use.

Three-dimensional observation of capillary nets in tumors

The rhodamine-labeled gelatin solution was dissolved at 37°C and loaded into a syringe connected to an 18 G needle. Under pentobarbital anesthesia, each animal's chest cavity was opened, and the right atrium and vena cava were incised for bleeding. The rhodamine-labeled gelatin solution was then slowly injected into the left ventricle until the circulated gelatin exited from the right atrium. Meticulous care was taken not to exert too much pressure while injecting. After perfusion, the animal was immediately immersed in an ice-cold fixative consisting of 0.5% paraformaldehyde (PFA) and 15% (v/v) aqueous saturated picric acid solution in 0.1 M Na-Phosphate buffer (pH 7.0) until the rhodamine-labeled solution gelled. The tumor was dissected along with the dorsal skin and observed under a fluorescent stereomicroscope, MZFLIII (Leica, Nassloch, Germany). The specimen was then postfixed with the same fixative at 4°C and kept in the dark. After washing the PFA away with PBS, the tumor was frozen on the cold stage and then sliced at a thickness of 200 μm with a sliding microtome. The slices were further fixed in 4% PFA in PBS overnight. After rinsing with PBS, each slice was mounted with mounting fluid consisting 0.05 M Tris-HCl-buffered saline (pH 8.0) containing 90% (v/v) of nonfluorescent glycerin and 10 mg/ml of 1,4-diazabicyclo-[2.2.2.]-octane (DABCO; Wako Pure Chemicals, Osaka, Japan). Each specimen was observed with a confocal laser scanning microscope (LSM-510; Carl Zeiss, Jena, Germany). After observation, some of the sections were dehydrated through a graded series of ethyl alcohol, embedded in an acryl resin (Teknovit 8100; Kulzer, Wehrheim, Germany), sectioned at a 5 μm thickness with a rotary microtome and observed after staining with methylene blue and basic fuchsin.

Quantitative analysis of development of capillary nets in tumors

After cardiac perfusion of the rhodamine-labeled gelatin, the 7-day tumors grown in both genotypes of mice (n = 5 each genotype) were dissected out, frozen and sliced at a 200 μm thickness. Twenty optical planes at every 10 μm from each slice were made by a confocal laser scanning microscope (objective lens × 10; 512 × 512 pixels = 1,000 × 1,000 μm). One-quarter areas of the original images were chosen from wherever an intratumor blood vessel could be observed. Each image size was then converted to 256 × 256 pixels, and the number of pixels in that area was counted using the software LIA for Win32. The tumor appeared as green pixels by GFP, and blood vessels showed red by rhodamine, while the overlapping area appeared in yellow. Therefore, the total numbers of yellow and red pixels were counted and became the vascular quantity. The numbers were statistically analyzed by Student's t-test using StatView software (SAS Institute, Cary, NC).

Quantitative analysis of protein by ELISA

The mice having the tumors were killed by an overdose of ether anesthesia. Subsequently, the tumor in each mouse was carefully dissected out, immediately placed in a 50-fold volume of extraction buffer, and then homogenized with a Polytron homogenizer. The extraction buffer was 50 mM Tris-HCl (pH 7.6), containing urea (1 M), NaCl (0.3 M), phenylmethylsulfonylfluoride (1 mM), eupeptin (1 mg/ml) and pepstatin (1 μg/ml). The homogenate was centrifuged for 10 min at 13,000g, and the supernatant was used as the protein solution for ELISA. Total protein concentration in each solution was determined by the BCA Protein Assay Reagent (Pierce, Rockford, IL).

Both human TN-C and human VEGF were measured using the conventional sandwich ELISA methods. For the detection of human TN-C, rat monoclonal antihuman TN-C antibody (7-13) was used as the first antibody, and biotinylated rat monoclonal antihuman TN-C antibody (36-13-6) as the secondary antibody. For the detection of human VEGF, rabbit antihuman VEGF antibody (Sc-507) was used as the first antibody; for the secondary antibody, biotinylated rabbit antihuman VEGF antibody (Sc-152) was used. The expression patterns of TN-C and VEGF were statistically analyzed by 2-way ANOVA using StatView software.

Coculture experiments

Both genotypes of fetal mesenchyme, TN−/− and TN+/+, were prepared from the fetuses of both on day 14 of gestation. The mesenchymal cells were then plated into the wells of 6 multiwell plates precoated or noncoated with mouse-TN-C. For coating the plate, mouse-TN-C was used at 5 μg/ml concentration. For the separation culture, the mesenchymal cells from TN+/+ fetuses were plated into the Transwell 3413 (Transwell, Corning Costar, Cambridge, MA) and cultured. To prepare the feeder cells, when the mesenchymal cells grew to the 80% confluence, mesenchyme growth was arrested by treatment with mitomycin (10 μg/ml) for 2 hr. After the plates were sufficiently washed several times with PBS, 3 × 104 A375-GFP cells were put on the feeder cells in each well. For the separation culture, 5 × 103 cells were directly plated into the noncoated wells of 24 multiwell culture plates, and the Transwell with the mesenchyme was then inserted into each well. The growth curves of A375-GFP cells in each well were obtained using the Premix WST-1 Cell Proliferation Assay System (succinate-tetrazolium reductase method; TaKaRa, Otsu, Japan). The solution colored by the enzyme reaction was measured at 450 nm, and 600 nm was used for the reference. The concentrations of both TN-C and VEGF under each experimental condition were measured on days 1, 3, 5 and 7 by ELISA as described above. The growth rates and concentrations of both proteins, TN-C and VEGF, were statistically analyzed by ANOVA (StatView software).

RESULTS

Tumor growth and development of capillary nets in tumors

The implanted A375-GFP cells had gradually grown to form a solid tumor in both WT and TNKO mice. At the first week after inoculation, there was no difference in the growth of tumors between the 2 animal groups, but by 3 weeks after transplantation, the WT mouse tumors had become significantly larger than those in TNKO mice (p < 0.0001; Fig. 1) and weighed twice as much. We hypothesized that this major difference in volume between the TNKO and WT tumors may have been due to some critical event that occurred during the early stage of tumorigenesis. Therefore, we focused our attention on the first stage of tumor growth.

Figure 1.

Transplantation experiment of A375-GFP into BALB/cA-nu TN+/+ (WT) and BALB/cA-nu TN−/− (TNKO) mice. The tumor in WT mouse was significantly heavier than that in TNKO mouse. After 3 weeks, the tumor in WT mouse increased in weight to approximately twice that in TNKO (p < 0.05). WT (open bar): wild-type BALB/cA-nude mice. TNKO (closed bar): TN-C gene knockout BALB/cA-nude congenic mice (n = 5). We performed 3 sets of experiments independently.

On day 1 after transplantation, the A375-GFP cells in WT mice had formed an aggregate in the subcutaneous tissue of each host mouse. This aggregate emitted the characteristic intense green fluorescence of GFP (Fig. 2a). A few blood vessels emitting red fluorescence of rhodamine had developed around the aggregate. The A375-GFP cells at the periphery of the aggregate were observed to emit this intense fluorescence, whereas no fluorescence was seen in the center (Fig. 2b). Newly formed blood vessels grew from the preexisting blood vessels but did not invade the aggregate (Fig. 2b).

Figure 2.

Vascular networks in the tumor. Rhodamine-labeled gelatin was perfused in the blood vessels of mice that received subcutaneous transplantations of A375-GFP cells. One (a, b), 3 (c, d), 5 (e, f) and 7 (g, h) days after transplantation in WT and 7 (i, j) days in TNKO, the tumor mass along with the skin was dissected and observed under a fluorescence stereomicroscope (a, c, e, g, i). The specimen was then sliced and observed under a confocal laser scanning microscope, and the projection images (b, d, f, h, j) were constructed by projecting 20 confocal images taken at an interval of 10 μm. Green fluorescence indicates the A375-GFP cell and red fluorescence the blood vessels in each image. Bar = 2 mm in (a), (c), (e), (g) and (i), and 200 μm in (b), (d), (f), (h) and (j). The A375-GFP cells formed an aggregate 1 day after transplantation. The aggregate as a whole emitted an intense green fluorescence, but no fluorescent cells are observed at its center (a, b). No blood vessels are observed in the aggregate (b). The fluorescent tumor cell aggregate is clearly demarcated 3 days after transplantation (c). Dense capillary nets surround the aggregate, which a few vessels penetrate (d). Most of the aggregate loses fluorescence but a small fluorescent mass of A375-GFP cells appears on day 5 (e). In the fluorescent mass of A375-GFP cells, a well-developed vascular network is formed (f). The mass grows on day 7 (g). The mass has an abundant vascular plexus (h). The fluorescent tumor cell mass formed in TNKO mice 7 days after transplantation has few vessels in comparison to that in WT (i, j).

As seen in the histologic sections of the aggregate, the A375-GFP cells in the center of the tumor were mostly damaged or necrotic (Fig. 3ac). The center was occupied almost entirely by numerous infiltrated polymorphonuclear cells (Fig. 3c). However, some healthy A375-GFP cells were present on the periphery of the aggregate and formed small cell masses (Fig. 3b). On day 3 after inoculation, the newly formed blood vessels increased in number. They were winding and surrounding the aggregate (Fig. 2c). The branching of blood vessels became more complicated, and some of them penetrated the tumor, but no blood vessels could be seen in the center of the aggregate yet (Fig. 2c and d). Only cells on the periphery emitted the GFP fluorescence as seen on day 1 (Fig. 2d). The polymorphonuclear cells were gradually disappearing from the aggregate with time (Fig. 3d and f). By day 5, the fluorescence in the primary aggregate was almost entirely gone, and the secondary tumor that emitted the intense fluorescence appeared beside the primary aggregate (Fig. 2e). As seen in Figure 2(f), a well-developed vascular plexus was formed in the secondary tumor. The tumor was composed of small lobules of A375-GFP cells separated by the connective tissue that was composed of fibroblasts and capillaries (Fig. 3e). The tumor with the GFP fluorescence had grown much larger by day 7 (Fig. 2g). The size of each lobule of A375-GFP cells had almost doubled by day 5 (Fig. 3g). Each cell within a lobule was tightly attached (Fig. 3f and g).

Figure 3.

Histologic observation of the tumor in WT mouse. Slices of the tumor shown in Figure 2 were further embedded in a plastic resin, cut at a thickness of 5 μm and observed under a light microscope after methylene blue and basic fuchsin staining. Low-power micrograph (a) and high-power micrographs (b, c) on day 1. High-power on day 3 (d) and day 5 (e). Low-power (f) and high-power (g) on day 7. Number on the bar indicates μm. The tumor cells form an aggregate on day 1 (a). At the periphery of the aggregate, numerous tumor cells having large nuclei with distinct nucleoli (arrows) are conspicuous (b), while tumor cells with a pyknotic nucleus (arrow heads), cell debris and polymorphonuclear cells dominate the center of the aggregate (c). The surviving tumor cells proliferate and form small assemblies on day 3 (d). In the tumor mass observed in Figure 2(e), tumor cells are assembled in lobules surrounded by connective tissue elements (e). The tumor mass on day 7 is composed of many lobules of tumor cells, which are tightly attached to each other, and capillary-like lumens (arrows) are remarkable between and within the lobules (f, g).

On the other hand, the A375-GFP cells transplanted into the TNKO mice underwent a development similar to that observed in WT mice during the first 5 days (Figs. 3e and 4e). However, although the capillary nets of the tumors in WT mice were well developed by day 7 (Fig. 2g and h), those in TNKO mice were scanty, lacking finer capillary nets (Fig. 2i and j). Furthermore, the histology of the tumors in TNKO mice (Fig. 3f and g) was quite different from that in the tumors of WT mice (Figs. 3f and g and 4f and g) in that it was markedly more fibrous (Fig. 4f), and each of the A375-GFP cells in the lobules was loose and its cytoplasm was atrophic (Fig. 4g).

Figure 4.

Histologic observation of the tumor in TNKO mouse. Slices of the tumor in TNKO mouse were further embedded in a plastic resin, cut at a thickness of 5 μm and observed under a light microscope after methylene blue and basic fuchsin staining. Low-power micrograph (a) and high-power micrographs (b, c) on day 1. High-power on day 3 (d) and day 5 (e). Low-power (f) and high-power (g) on day 7. Number on the bar indicates μm. The histologic observations of the tumor in TNKO mouse closely resemble those in WT mouse on day 1 to day 5, but diverge widely on day 7. The tumor cells in the mass are assembled in lobules or cords, and the connective tissues between them are relatively abundant (f). The tumor cells in the lobules are comparatively reduced in size and loosely attached to each other (g).

Quantitative analysis of development of capillary nets in tumors

The amount of blood vessels in the tumor of a WT mouse was 106780.4 ± 19483.738 pixels, whereas that in the tumor of a TNKO mouse was only 21907.8 ± 5442.96. Thus, the degree of neovascularization in WT mice was significantly higher than that in TNKO mice (p < 0.003).

Quantitative analysis of expression of TN-C and VEGF in vivo

Both human TN-C and human VEGF in the tumors were measured by the sandwich ELISA method. The TN-C content in the tumor grown in WT mouse remained unchanged until day 3, after which it significantly increased up to day 7 (Fig. 5a, open circle). In contrast, TN-C in TNKO mice rapidly increased up to day 5 when it reached its highest level, steadily decreasing thereafter (Fig. 5a, closed circle). Interestingly, the amount of TN-C in the TNKO mouse tumors was significantly higher than that in WT mice from day 3 up to day 5 (Fig. 5a), with the respective p-values of < 0.0255 and < 0.0001.

Figure 5.

Expression profiles of TN-C and VEGF in the tumor inoculated in TNKO and WT mice. (a) Changes in the amount of TN-C in the total protein (μg/mg) during tumor growth. TN-C content in the tumor grown in the WT mouse is stable between day 1 and day 3 after implantation and gradually increased from days 3 to 7 (open circle). In contrast, in the TNKO mouse it is most abundant 5 days after implantation, rapidly increases between 3 and 5 days and then decreases thereafter (closed circle). By 5 days, the implant in the TNKO mouse contains significantly higher concentrations of TN-C than the one in the WT mouse. (b) Changes in the amount of VEGF/total of protein (μg/mg) in the tumor. VEGF content in the tumor in the TNKO mouse is low and roughly stable during the experimental period (closed circle), while that in the WT mouse exhibits a transient and significant rise 5 days after implantation (open circle) but is almost comparable to the former on other days studied.

On the other hand, the content of VEGF in the tumors grown in WT mice was at its highest on day 5 (Fig. 5b, open circle), although there had been no difference in the contents between WT mice and TNKO mice by day 3. However, this amount in TNKO mice remained low and roughly stable during the transplantation period (Fig. 5b, closed circle). This tendency was also seen in further experiments performed on another day.

Effect of embryonic mesenchyme on A375-GFP cell growth

The growth of A375-GFP cells was suppressed by the cocultivation with embryonic mesenchyme taken from day 14 fetuses. Therefore, in order to determine how the mesenchyme suppressed the growth of A375-GFP, we used the Transwell to separate the embryonic mesenchyme from A375-GFP. In this culture, we can determine whether a certain factor, if any, may affect the growth of A375-GFP without cell-cell contact.

Our results showed that the embryonic mesenchyme cultured in the Transwell arrangement secreted a certain diffusible factor, which then inhibited the growth of A375-GFP cells (data not shown). As seen in Figure 6, the growth of A375-GFP cells was reduced when it was cocultured with both genotypes of embryonic mesenchyme involving an intermingling of both cells (Fig. 6, open square, open triangle, closed square, closed triangle). In contrast, the A375-GFP cells cultured alone grew well on the noncoated plastic culture dish (Fig. 6, open circle). By day 3, the A375-GFP cells with the TNKO mesenchyme (Fig. 6, open triangle) grew as well as those with WT mesenchyme (Fig. 6, open square), but that growth subsequently slowed down. There were significant differences between TNKO and WT on day 5 and day 7 (p < 0.0004 and < 0.0001, respectively). However, this deficiency in the TNKO mesenchyme was corrected by additional TN-C, with cell growth achieving the same level as cells with WT mesenchyme (Fig. 6, closed triangle). However, although this additional TN-C slightly enhanced the growth of A375-GFP cultured alone by day 5 (Fig. 6, closed circle), it failed to enhance the function of WT mesenchyme (Fig. 6, closed square).

Figure 6.

A375-GFP cell growth curve in the cocultivation with embryonic mesenchyme. The growth of A375-GFP cells is reduced when it was cocultured with both genotypes of embryonic mesenchyme involving an intermingling of both cells (open square, open triangle, closed square, closed triangle). In contrast, the A375-GFP cells cultured alone grow well on the noncoated plastic culture dish (open circle). By day 3, the cell growth with TNKO mesenchyme is rather well advanced in comparison to that with WT mesenchyme (p < 0.0305, open square and open triangle). Subsequently, the A375-GFP with WT mesenchyme is gradually growing by day 5, but that with TNKO mesenchyme is not (p < 0.0004, open square and open triangle at D5). Finally, by day 7 there is a significant difference in cell growth between the A375-GFP with WT mesenchyme and that with TNKO mesenchyme (p < 0.0001, open square and open triangle at D7). Note that additional TN-C can correct the deficiency of the TNKO mesenchyme described above (closed triangles). Under the existence of additional TN-C, A375-GFP cells with TNKO mesenchyme show a similar growth curve to cells with WT mesenchyme. However, although this additional TN-C slightly enhances the growth of A375-GFP cultured alone by day 5 (closed circle), it fails to enhance the function of WT mesenchyme (closed square).

Effects of embryonic mesenchyme on expression of both TN-C and VEGF in cocultivation

To analyze the effect of embryonic mesenchyme on the expression of TN-C and VEGF in A375-GFP cells, we measured both human TN-C and human VEGF in the culture media of A375-GFP cells by sandwich ELISA. Our rat monoclonal antihuman TN-C antibody never detected the mouse TN-C secreted by WT mesenchyme because it is a human-specific antibody; and TNKO mesenchyme, of course, never secreted TN-C (data not shown). The expression of TN-C in A375-GFP cultured alone had increased by day 3 and remained roughly stable up to day 7 (Fig. 7a, open circle). Interestingly, there was no effect of either genotype of the mesenchyme on the expression of TN-C by day 3 such as was seen in the growth of A375-GFP cells. The WT mesenchyme then induced more TN-C in A375-GFP cells (Fig. 7a, open square) than the TNKO mesenchyme (Fig. 7a, open triangle) from day 5 up to day 7. However, the TNKO mesenchyme did not induce TN-C in A375-GFP, but rather suppressed it (Fig. 7a, open triangle).

Figure 7.

Expression profiles of TN-C and VEGF in the cocultivation between A375-GFP and embryonic mesenchyme. (a) Changes in the amount of human TN-C/total protein (μg/mg) in A375-GFP cells cocultured with embryonic mesenchyme. The expression of TN-C from the A375-GFP cultured on plastic increase by day 3 and stays at a stable level for culture periods (open circle). WT mesenchyme enhances TN-C expression (open square). In contrast, when A375-GFP cells were cocultured with TNKO mesenchyme, TN-C expression is downregulated (open triangle). However, additional TN-C coated onto a culture dish reverses this deficiency and raises the TN-C expression levels by the same level as that with WT mesenchyme (closed triangle). Additional TN-C has no observable effect on either the level of TN-C expression in cocultivation with WT (closed square) or the one in A375-GFP cultured on plastic (closed circle). (b) Changes in the amount of VEGF/total protein (μg/mg) in the tumor. VEGF in A375-GFP cultured on plastic decreases in amount by day 5 and increases by day 7 (open circle). However, VEGF in the A375-GFP cocultured with WT mesenchyme shows a higher level even on day 1 and rapidly reaches its peak by day 3; afterward, its level rapidly decreases and returns to control levels by day 7 (open square). In contrast, TNKO mesenchyme does not show any stimulatory effect on VEGF expression (open triangle). However, this ineffectiveness of TNKO mesenchyme is restored by additional TN-C, although the expression is remained at an intermediate level (closed triangle). Additional TN-C does not show any effect on VEGF expression in A375-GFP cocultured with WT mesenchyme (closed square) or the expression in A375-GFP cultured on plastic (closed circle).

Surprisingly, the additional mouse TN-C restored the suppression of TN-C expression (Fig. 7a, closed triangle), so that the expression of TN-C in this case took the same course as that in the WT mesenchyme (Fig. 7a, open square). Our study proved that the additional mouse TN-C had no direct effect on the expression of TN-C in A375-GFP cells (Fig. 7a, open circle vs. closed circle), and that the effect of the WT mesenchyme was not augmented by the additional TN-C (Fig. 7a, closed square).

Furthermore, to verify the molecular mechanisms of angiogenesis during tumorigenesis, the sandwich ELISA method was used to analyze the regulation of human VEGF expression. In this study, we used 2 antihuman VEGF antibodies.

The amount of VEGF secreted by A375-GFP had initially decreased until day 5, and then returned to its original level by day 7 (Fig. 7b, open circle). However, although the WT mesenchyme strongly induced the expression of VEGF in A375-GFP until day 3, its level rapidly decreased thereafter, returning to control levels by day 7 (Fig. 7b, open square). In contrast, the TNKO mesenchyme showed no inductive effect on the expression of VEGF (Fig. 7b, open triangle), but instead seemed to inhibit it somewhat in A375-GFP. Nevertheless, this inhibitory effect was improved by the additional TN-C, although the recovery level of VEGF expression remained at an intermediate level (Fig. 7b, closed triangle). Interestingly, the additional TN-C had absolutely no effect on the expression of VEGF in A375-GFP cultured alone (Fig. 7b, closed circle) and failed to enhance the inductive effect of WT mesenchyme (Fig. 7b, closed square).

Taken together, these data clearly indicated that the WT mesenchyme has a potential for the induction of both TN-C and VEGF expression in A375-GFP, but that the TNKO mesenchyme had no such inductive effect.

DISCUSSION

In this transplant experiment, a human melanoma cell line (A375-GFP cells) formed a solid tumor underneath the nude mouse skin. However, the growth of tumors in TNKO mice was suppressed due to the lack of TN-C from a stromal microenvironment (Fig. 1). Therefore, to determine the cause of this suppression of TNKO mouse tumors, we performed not only 3-dimensional observations of the capillary nets in the tumors but also a quantitative analysis of their development. As described above, these analyses revealed obvious differences in angiogenesis between the tumors in WT mice and those in TNKO mice. In particular, the quantitative analysis clearly indicated that the networks of newly formed capillaries in a WT tumor were more abundant than those in TNKO mice.

As for the angiogenesis, VEGF is the only growth factor that has been observed almost ubiquitously at sites of angiogenesis and whose levels correlate most closely with the spatial and temporal stages of blood vessel growth. It has been well established that the inhibition of VEGF activity results in the growth suppression of a wide variety of tumor cell lines in murine models,32, 33i.e., the amount of VEGF seems to correlate closely with the degree of angiogenesis. Therefore, we focused on VEGF as a key molecule in angiogenesis. Indeed, the content of VEGF in TNKO mouse tumors was lower than that in WT mice (Fig. 5b). Meanwhile, the content of VEGF in WT mouse tumors was highly abundant only on day 5, but no obvious difference between WT and TNKO was observed on any other days. Contrary to our expectations, however, the content of human TN-C in TNKO mouse tumors was much greater than that in WT mice from day 3 to day 5, although it rapidly decreased thereafter (Fig. 5a). The reason the TN-C expression of TNKO mouse tumors was higher than that in WT mice seems to be due to the lack of cell-cell interaction mediated by some mesenchymal factor.

There are 2 different types of cancerous cell lines: one can produce TN-C autonomously under cultivation, the other cannot.34 Interestingly, most cancerous cell lines begin to express TN-C when they are transplanted beneath the nude mice skin, forming solid tumors.11, 16 Furthermore, when TN-C nonproducing culture cells were transplanted subcutaneously in the syngeneic TNKO mouse, they did not express TN-C, although they formed a solid tumor.16 It was also reported that the expression of TN-C in the tumor was induced by a certain diffusible factor in the embryonic mesenchyme.11 This, a TN-C inducing factor, as it is known, is also involved in tumor growth during lung metastasis.35

The expression of the TN-C of transplanted cancerous cells probably depends on a certain regulation of the stromal environment. Therefore, in this study, it was not unreasonable to think that the expression of TN-C in A375-GFP may subsequently come to depend on some mesenchymal factor, even though A375-GFP can autonomously produce TN-C by day 5. In other words, when A375-GFP was transplanted underneath the skin of WT mice, its TN-C expression seemed to be controlled by the factor secreted by the surrounding mesenchyme. Consequently, it is presumed that A375-GFP continued to produce TN-C autonomously up to day 5, though they required mesenchymal support thereafter.

These findings clearly demonstrated that the expression of VEGF is regulated by mesenchymal TN-C. Moreover, it is worth noting that there is a critical day for the onset of cell-cell interactions that are involved in the induction of both TN-C and VEGF expressions in the tumor. We could speculate that guidance control of the mesenchyme over the VEGF expression of A375-GFP might be mediated by some mesenchymal factor whose expression could be regulated by its own TN-C. However, we were unable to elucidate why the expressions of both TN-C and VEGF reached a peak on day 5 after transplantation. Therefore, we performed the cocultivation in order to examine the involvement of the surrounding mesenchyme in the expression of both TN-C and VEGF in A375-GFP as well as the function of TN-C of the mesenchyme in cell-cell interactions.

In this cocultivation, we used the embryonic mesenchyme instead of the matrix cells of the tumor. When A375-GFP was cultured under mixed conditions, its growth was suppressed by the embryonic mesenchyme. We initially thought that this suppression was due to a nutrition problem. Therefore, we used the Transwell to separate the 2 cellular components. Nevertheless, the growth of A375-GFP was suppressed by a probable diffusible factor from the mesenchyme. Interestingly, the degree of the cell growth suppression from the mesenchyme of WT mice was similar to that in TNKO mice by day 3 of culture. Eventually, the mesenchyme of the TNKO mouse strongly suppressed the cell growth. However, the additional TN-C corrected the defect in the TNKO mouse mesenchyme. It is worth noting that this exogenous TN-C influenced neither the growth of A375-GFP nor the supportive ability of the WT mouse mesenchyme. Thus, these findings clearly indicated that the onset of cell-cell interactions between cancerous cells and mesenchymal cells did not begin by day 3. Furthermore, it was clear that the additional TN-C acted on the mesenchyme, not on the A375-GFP.

As to the expressions of both TN-C and VEGF in cocultivation, A375-GFP showed a different response to the mesenchyme. In the expression of TN-C, there was no difference among the cocultivation cases by day 3. However, the WT mouse mesenchyme increased the expression of TN-C in A375-GFP up to day 7, whereas the TNKO mouse mesenchyme rather inhibited their expression after day 3. The additional TN-C again corrected the defect in the TNKO mouse mesenchyme, and the level of TN-C expression reached that of the WT mouse mesenchyme. The expression level of VEGF was also corrected, but it remained at an intermediate level. It was also noted that the additional TN-C showed no effect on the expression of either the TN-C or VEGF of A375-GFP cultured alone. Consequently, the additional TN-C seemed to influence the expression of these molecules through the mesenchyme, rather than the A375-GFP directly. In the expression of TN-C in A375-GFP, the cell-cell interactions began after day 3.

On the other hand, the WT mouse mesenchyme enhanced the expression of VEGF, but then declined after day 3, and finally no longer functioned on day 7. This would appear to be similar to the in vivo data. However, it provides no answer as to why the regulation of VEGF expression by the mesenchyme was temporary.

In summary, the present experiments revealed the following. One, mesenchymal TN-C is involved in cell-cell communications between cancerous and mesenchymal cells. Two, mesenchymal TN-C is particularly indispensable to the normal function of mesenchyme. Three, mesenchymal TN-C is especially important for inducing some diffusible factors that allow cancerous cells to produce both TN-C and VEGF. Four, the onset of the cell-cell interactions between cancerous and mesenchymal cells needs more than 3 days after encountering these cells. Five, day 5 may be the critical day for the cell-cell interactions mediated by both in vivo and in vitro TN-C. Six, the expressions of TN-C and VEGF seem to be regulated by different molecular mechanisms, since the cellular response of the A375-GFP cells differed between them. In conclusion, these results suggest that it is important for cancerous cells to interact with the host mesenchymal cells, and that mesenchymal TN-C will play an important role not only in the mesenchymal function but also in the more advanced stage of angiogenesis.

Acknowledgements

The authors thank Prof. Hiroshi Ishikawa, Department of Anatomy, Jikei University School of Medicine, for his valuable discussions and encouragement throughout this work, which was supported in part by a grant to M.K. from Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

Ancillary