Luminal architectures with dendritic branching are widely observed in various structures in life, such as the bronchi, blood vessels, and exocrine glands in animals and leaf veins in dicotyledonous plants. In the vascular tissues of mammals, it has been reported that cell growth-related molecules, adhesion molecules, and extracellular matrix proteins are involved in dendritic branching formation (Fisher et al., 2001; Davies, 2002; Hoffman et al., 2002; Sakai et al., 2003). However, it is not yet well understood how these molecules determine the shape of vasculature (vascular morphogenesis) in various organs.
Developmental processes of the vascular system have been extensively studied. Two distinct mechanisms, vasculogenesis and angiogenesis, are required for vessel formation in the embryo (Carmeliet, 2000; Yancopoulos et al., 2000). Early in development, vessel formation occurs by a process called vasculogenesis, in which endothelial cells within a vascular tissue differentiate in situ and coalesce to form a primary capillary plexus. The primary capillary plexus develops into many important vessels, including the heart, aorta, and vessels in the retina and the mesentery. The primary capillary plexus then undergoes remodeling with angiogenesis (Dunlop et al., 1997; Carmeliet, 2000). The remodeling required for vascular morphogenesis is achieved via sprouting and regression of the capillaries. Finally, the capillaries develop into mature vessels with the recruitment of pericytes and smooth muscle cells. Consequently, the regulation of remodeling processes is considered to be crucial for vascular morphogenesis.
An examination of the phenotypes of mutant embryos lacking stromal cell-derived factor-1 (SDF-1) or its receptor, CXCR-4, showed that the factor as well as its receptor play essential roles for remodeling of the primary capillary plexus in particular organs (Tachibana et al., 1998). In these phenotypes, vascular anomalies due to the loss of remodeling were found in the alimentary tracts, though no vascular anomalies were observed in any other organs. These findings suggest the possibility that organ-specific molecular factors play key roles in angiogenesis in various organs. Therefore, in vivo studies are required for the elucidation of such molecular factors in the vascular morphogenesis of each specific organ.
Del1 is an embryonic extracellular matrix protein, which has an RGD motif and binds to integrin αvβ3 and αvβ5 (Hidai et al., 1998; Zhong et al., 2003). Due to its expression pattern and its role in promoting vascularization in in vitro studies (Hidai et al., 1998), Del1 has been suggested to be involved in vascular development. However, Del1 function in embryos has not been studied. In the present study, to investigate the role of Del1 in embryonic development in vivo, we have generated mice that overexpress Del1. The transgenic mutant mice exhibited more sophisticated dendritic branching than wild-type mice. The observation suggests the existence of a novel mechanism for vascular branching morphogenesis.
To evaluate the role of Del1 during embryogenesis, transgenic mice that overexpressed Del1 during gestation were generated. The animals were produced using an expression vector, pCAGGS, containing the cytomegalovirus enhancer sequence (Fig. 1A) (Niwa et al., 1991). In total, six lines of transgenic mice were produced. Three lines (C35, C49, and C50) out of the six, all of which grew and reproduced normally according to Mendel's law, were selected for further investigations. Genotyping of each mouse was examined by tail blotting, and the expression of the transgene was confirmed by Northern blot analysis (Fig. 1B).
We performed macroscopic screening of the vascular system in transgenic mice; however, no vascular anomalies were found in major organs and tissues other than the mesentery (data not shown). The morphological differences from wild-type mice were found in mesenteric vessels from the gastric side of the jejunum, which was defined as the region of interest. Arteries and veins were paired in the mesenteries of both wild-type and transgenic mice, indicating that overexpression of Del1 did not affect the differentiation of the artery or the vein (Fig. 2A). Since mesenteric veins were much clearer and more distinct than mesenteric arteries, morphological analysis using the veins enabled accurate quantification of morphological characteristics. Therefore, we employed veins as the target for morphological analysis. Several vessels arising from the root circulated the region of interest. The wild-type mesenteric veins typically had a rake-like appearance, consisting of a long vessel and several short branches at its end (Fig. 2B). In contrast, the typical transgenic veins had a general tree-like appearance (Fig. 2C–F).
There were some differences in the appearance of mesenteric vessels among the transgenic lines examined (Fig. 2C–F). To investigate the relationship between expression levels of the transgene and morphological changes, Northern blot analysis was performed. As shown in Figure 3A, the expression levels of the transgene in mesentery differed between the lines examined. The transcripts from the endogenous Del1 were not detectable in the mesentery of adult wild-type mice. The expression level of exogenous Del1 was low in C49, medium in C50, and high in C35.
The mesenteric vessels of wild-type and transgenic mice showed the genotype-specific patterns in which certain individual differences in patterns were additionally involved as shown in Figure 2E and F. Therefore, statistical analysis was required to elucidate genotype-specific differences in vascular morphology between wild-type and transgenic mice. In our statistical analysis, we focused on two parameters: the distance from the root of the mesentery to the first branching point and the total length of veins including branches. The former might reflect the effect of Del1 on branching morphogenesis, while the latter might confirm whether Del1 is angiogenic or angiostatic.
As shown in Figure 3B, the distance from the root of the mesentery to the first branching point was 165 ± 8 pixels in wild-type mice, 118 ± 6 pixels in C49, 126 ± 6 pixels in C50, and 78 ± 3 pixels in C35. As expected from the impression obtained from Figure 2, branching patterns of mesenteric veins were certainly affected in transgenic mice.
As shown in Figure 3C, the total vessel length was 1,990 ± 45 pixels in wild-type mice, 1,601 ± 28 pixels in C49 (80% of wild-type mice), 1,506 ± 32 pixels in C50 (76%), and 1,339 ± 34 pixels in C35 (68%). The total length of veins in transgenic mice was significantly shorter than that in wild-type mice. Both the distance from the root of the mesentery to the first branching point and the total length of veins including branches decreased in inverse proportion to the expression level of Del1 mRNA. This altered first branching position in vessels seems to be responsible for producing the dendritic appearance in transgenic animals.
To examine whether the morphological changes described above were secondary effects due to abnormal development of the intestine induced by Del1 overexpression, morphological analysis of the jejunum and chronological observation of body weight gain were undertaken. We measured the length of the region of interest of the jejunum. There was no significant difference between 6.1 ± 0.1 cm in wild-type mice and 6.2 ± 0.1 cm in transgenic mice. The jejunum of transgenic mice was of normal structure in macroscopic (Fig. 2) and microscopic (data not shown) observations. Transgenic mice also grew steadily, as did wild-type mice (Fig. 3D). The body weight was 20.8 ± 0.4 g in wild-type mice and 20.3 ± 0.4 g in transgenic mice at 8 weeks after birth.
As described above, morphological changes in the vascular system were not found in other organs. We thoroughly examined vessels around the mesentery of the jejunum. Vessels on the surface of the jejunum were dendritic in wild-type and transgenic mice (Fig. 4A and B). The capillary of the jejunum showed a honeycomb pattern both in wild-type and transgenic mice (Fig. 4C and D). The mesenteric vessels in the ileum were analyzed morphologically. First, the expression of transgene in the mesentery of the ileum was examined. Overexpressed Del1 was detected in the mesentery of transgenic mice (Fig. 4E). Branching patterns in transgenic mice were not different from that in wild-type mice (Fig. 4F and G). The distance of the first branching point from the root and the total vascular length were not significantly different between wild-type and transgenic mice in the ileum (Fig. 4H and I).
Our data indicated that the vascular phenotype in Del1 transgenic mice was specific in the mesentery of the jejunum. In humans, branches of the mesenteric arteries are scarce in the jejunum and become increasingly more frequent in the ileum (Healy, 2005). This observation may suggest a specific developmental mechanism for the mesenteric vessels in the jejunum and the ileum. Consistent with this observation in humans, the measurement of intestinal length and mesenteric vascular length in wild-type mice revealed that the ratio of vascular length to intestinal length was less in the jejunum than that in the ileum (1.60 ± 0.10 vs. 1.86 ± 0.14; P < 0.01; n = 6).
To investigate the chronological change in the morphological pattern of mesenteric vessels in wild-type and transgenic embryos, whole mount immunohistochemistry with an antibody against PECAM-1 was performed on the alimentary tract of embryos at mid-gestation, which corresponds to the onset of vasculogenesis and remodeling progression (Fig. 5). We could not find any differences in the vasculature of either the intestines or the mesenteries of E10.5 wild-type and transgenic embryos (data not shown). Observations of the primary vascular network in both wild-type and transgenic mice showed that remodeling of the primary network had, at the E10.5 stage, not yet progressed in the mesentery. At E11.5, some differences in mesenteric vessels between wild-type and transgenic mice were observed (Fig. 5A–D). In wild-type mesentery, remodeling resulted in rake-like vessels with a long stalk (Fig. 5A and C). In contrast to wild-type mice, the vessels of transgenic mesentery were irregular in shape, luminal size, and distribution. Large branches from trunks were found only in the transgenic embryos (Fig. 5B and D).
Although any vascular phenotypes other than that in the mesentery could not be found in adult transgenic mice, the whole mount immunohistochemistry results for embryos revealed that the developmental process of gastric vessels was affected by Del1 overexpression. No differences could be found in the vasculature of the stomachs of wild-type and transgenic embryos at E11.5 (Fig. 5E and F). Components of the primary vascular network remained at the periphery of the stomach, although the remodeling into large vessels had progressed around the cardia. Stomachs from wild-type mice at E12.5 showed large vessels around the cardia and the presence of the primitive network at periphery (Fig. 5G). However, the stomachs of E12.5 transgenic embryos showed that remodeling of the primitive network had been almost completed, even in peripheral regions (Fig. 5H). Finally, there were no differences observed in vascular morphology of the stomach between adult wild-type and transgenic mice (data not shown). Such acceleration of vascular remodeling could not be found in any other vessels.
It has been reported that periendothelial cells play an important role in vascular morphogenesis by Ang-1 (Sato et al., 1995; Suri et al., 1996) and ephrin-B2 (Adams et al., 1998; Wang et al., 1998). In this study, electron microscopic observations demonstrated that the endothelial cells and the periendothelial cells in transgenic mesenteries were normal at E11.5 (Fig. 6).
Since the overexpression of the Del1 gene affected the morphogenesis of mesenteric vasculature, it was worthwhile to examine the spatial and temporal expression patterns of the endogenous Del1 gene in mouse embryos in order to investigate the role of Del1 in vascular morphogenesis of mesenteric vessels. In wild-type mice, the endogenous Del1 gene was expressed in the endothelial cells of the primary vascular plexus at 11.5 days postcoitum (E11.5; Fig. 7A and B). However, Del1 transcripts were not detected in mesenteric blood vessels at E16.5 (Fig. 7C). In contrast, mice overexpressing Del1 were shown to have abundant Del1 transcripts in the abdominal tissues, including the mesenteric blood vessels, at both E11.5 (data not shown) and E16.5 (Fig. 7D). We performed RT-PCR using RNA from the mesenteries and intestines of both wild-type and transgenic animals at various stages of development (Fig. 7E). Del1 was highly expressed at E11.5 and gradually declined after vascular remodeling in the mesenteries of wild-type mice. On the other hand, the exogenous Del1 gene in transgenic mice was continuously expressed at high levels.
Several genes have been found to be essential for normal vascular development, such as VEGF, Ang-1, and their receptors. To investigate whether overexpression of Del1 affects the expression of these genes, RT-PCR was performed for mRNA from the abdomen of E11.5 embryos. As shown in Figure 8, the expression level of Del1 in transgenic embryos was much higher than that in wild-type embryos. However, no significant differences between wild-type and transgenic mice were observed in the expression levels of SDF-1, CXCR-4, Ang-1, Tie-2, VEGF, and Flk-1 (Shalaby et al., 1995; Carmeliet et al., 1996; Suri et al., 1996).
Since the morphological changes of blood vessels in transgenic mice were localized specifically in mesenteric vessels, we hypothesized that Del1 could work downstream of SDF-1 and CXCR-4 signaling (Tachibana et al., 1998). To investigate the effects of SDF-1 on Del1 expression of endothelial cells, an in vitro experiment was performed with SDF-1 in the culture medium. As shown in Figure 9, Del1 mRNA from human umbilical vein endothelial cells (HUVECs) increased, peaking at 18 hr of SDF-1 administration.
In this study, neither macro- nor microscopic abnormalities were found in the stomach or intestine of transgenic mice. Transgenic mice gained body weight normally after birth (Fig. 3D). Therefore, vascular phenotypes in transgenic mice are not attributed to impaired alimentary function, but probably occur through direct effects of Del1 overexpression. To our knowledge, a phenotype of this nature generated by gene engineering has not been previously reported.
It has been reported that Del1 has an RGD sequence in its secondary EGF repeat and binds to integrin αvβ3 and αvβ5 (Hidai et al., 1998; Zhong et al., 2003). Other extracellular matrix proteins such as fibronectin and fibrillin also have an RGD motif and bind to integrins (Hohenester, 2002). Since Sakai et al. (2003) have reported that fibronectin plays a pivotal role in epithelial branching, it is possible that the overexpressed RGD motif of Del1 occupies binding sites of integrin to other extracellular matrix proteins, which results in the dendritic branching in mice overexpressing Del1 (Friedlander et al., 1996; Bader et al., 1998; Reynolds et al., 2002). However, since the effects of overexpressed Del1 were only observed in the mesenteric vessels, this explanation is not entirely plausible. The expression pattern of the endogenous Del1 in the mesentery of embryos also suggested a possibility that the endogenous Del1 contributed to the development of mesenteric vessels (Fig. 7). Loss-of-function experiments are necessary to determine if Del1 is significant and/or essential in the normal development of the mesenteric vessels.
Morphological changes in vascular branching were restricted in the mesentery of the jejunum and embryonic stomach (Figs. 2, 4, and 5). In human, ileal branches are more numerous than the jejunal branches (Healy, 2005). In this study, it was confirmed that the rule is also valid in mice. Specific sensitivity of the jejunal mesenteric vessels to Del1 overexpression could be involved in the development of the anatomical characteristics.
The distribution of vascular morphological changes in Del1 transgenic mice suggests a close relationship between Del1 and the SDF-1/CXCR-4 signaling systems (Tachibana et al., 1998). Vascular anomalies are also localized in the alimentary tracts in SDF-1 null mutants, where the primary capillary plexus cannot undergo angiogenesis. The finding that Del1 overexpression affected the vascular morphogenesis only in the alimentary tracts, as well as the finding that the increasing administration of exogenous SDF-1 increased the expression of the Del1 gene in in vitro experiments (Fig. 9), suggests that SDF-1/CXCR-4 signaling regulates angiogenesis in the mesentery, mediated by the activation of the Del1 gene.
Gene targeting experiments have revealed that several secreted molecules and their receptors are essential for vascular remodeling in mouse embryos (Sato et al., 1995; Suri et al., 1996; Tachibana et al., 1998; Wang et al., 1998; Peng et al., 2000). The primary capillary plexus in mutant embryos that lack Angiopoietin-1 (Ang-1) or Tie-2 does not undergo normal angiogenesis because it fails to recruit pericytes (Sato et al., 1995; Suri et al., 1996). Ang-1 overexpression mice had larger, more numerous, and more highly branched vessels (Suri et al., 1998). Mouse embryos lacking ephrin-B2 and its receptor, EphB4, suffer fatal defects that are similar to those seen in mice lacking Ang-1 or Tie-2 (Adams et al., 1998; Wang et al., 1998). In the present study, the expression of classical angiogenesis-related genes was evaluated by conventional RT-PCR. It was not affected in transgenic mice (Fig. 8). Although conventional RT-PCR does not effectively provide precise quantification of gene expression, we suggest that the dendritic pattern in mesenteric vessels by Del1 overexpression probably occurred without massive modification of those genes known to be involved in remodeling.
To investigate the biological function of Del1, several assays have been employed. In the initial chicken chorioallantoic membrane (CAM) assays employing tumor cells expressing Del1, loss of vascular integrity and promotion of vascular remodeling were observed (Hidai et al., 1998). In contrast, another CAM assay experiment showed the purified recombinant Del1 protein was highly angiogenic (Penta et al., 1999). Furthermore, Del1 stimulates angiogenesis in explanted tumors (Aoka et al., 2002). In Del1 transgenic mice, the total vascular length was decreased in the mesentery (Fig. 3C). However, Del1 cannot be considered as a simple angiogenic inhibitor in developing embryos, since capillary vessels did not show any abnormality in the mesentery and the decreased total vascular length could be the secondary effect of the dendritic branching due to Del1 overexpression. In this study, it is noteworthy that dendritic branching was induced with a decrease of vascular length. The dendritic branching vasculature may be biologically resource-saving.
It was thought that the observed change in branching patterns from wild-type to transgenic mice would not be quantifiable, since it appeared to be a qualitative change. Novel analytical methods would be required to measure the vascular phenotype of Del1 transgenic mice. In the present study, the direct mechanism of dendritic branching could not be established. However, it should be mentioned that dendritic branching of alimentary vessels was induced by a single gene modification without any sophisticated local regulation of gene expression. The phenotype suggests the presence of a new mechanism for determining branching patterns in the mesentery.
Recently, molecules associated with vascular remodeling have been considered as targets for clinical applications. Such molecules, in conjunction with other angiogenic factors, are expected to form better blood vessels with many branches into ischemic regions (Asahara et al., 1998; Peters, 1998; Holash et al., 1999). Further investigations of extracellular matrix proteins in vascular development may give us new insights into potential clinical applications.
Generation and genotyping of transgenic mice.
The plasmid construct used for generating the transgenic lines in this study contained 2.0 kb of the mouse Del1 cDNA, including the entire coding region. The sequence was cloned into an EcoRI site in the expression vector pCAGGS (a kind gift from Dr. Miyazaki, Osaka, Japan), which included the cytomegalovirus enhancer sequence (Fig. 1A) (Niwa et al., 1991). Digestion of the construct with SalI and BamHI generated a 4,460 bp fragment for injection. Southern blotting analysis was used to screen founders and genotype each mouse (Fig. 1B). For Southern blot analysis, 10 μg of genomic DNA, extracted from either the tails or placentas of mice, was digested with EcoRI, size-fractionated on 1% agarose gels, transferred to nitrocellulose, and finally hybridized to a 2.0 kb fragment corresponding to Del1 cDNA as found in the transgene. The Alphosdirect system (Amersham Pharmacia) was employed for probe-labeling, hybridization, and detection.
Northern blot analysis and RT-PCR.
The mesentery dissected from adult mice and embryos was homogenized with a polytron, and RNA was isolated with TRIzol (Sigma) according to the manufacturer's instructions. For Northern blot analysis, 20 μg of total RNA was size-fractionated on 1.3% agarose gels containing 2.2 M formaldehyde, transferred to nitrocellulose, and hybridized to the 2.0 kb cDNA fragment of the Del1 gene. For RT-PCR, 1 μg of total RNA was used for reverse-transcription reactions (Invitrogen). The sequence, the melting temperature (Tm), and the product length for PCR primers are listed in Table 1. For each reaction, the number of cycles was optimized to make it certain that the signal accumulated within the linear range. The PCR product was electrophoresed on a 2% agarose gels, transferred to nitrocellulose, and hybridized with internal oligonucleotides. The mouse S16 ribosomal protein (ms16) gene was employed as an internal control for the RT-PCR (Hidai et al., 2003). Experiments were repeated at least three times, and representative data are shown in the figures.
Table 1. Primers sequences for RT-PCR
5′ Primer (5′ to 3′)
3′ Primer (5′ to 3′)
Analysis for vessels in mesentery.
For anatomical analysis of major vessels of adult mice, an angiographical method with Indian ink was employed. After injection of Indian ink via the left ventricle, vasculature of the brain, neck, chest, lung, heart, kidney, and liver was carefully observed. For precise analysis of vessels in the mesentery and on the surface of the intestine, mice (7–8 weeks) were sacrificed by deep anesthesia. Then a small hole was made in the parietal peritoneum. A 1 mL aliquot of 4% paraformaldehyde was injected through this opening. After 30-min fixation, the intestine and the colon with the surrounding mesentery were excised and suspended in PBS to be supplied for macroscopic observation.
On the oral side of the jejunum, where 20 oral-side veins flow from the jejunum into the mesentery, vascular branching in transgenic mice was apparently different from that in wild-type mice. We therefore selected this region as the region of interest. Images of the samples were collected at a resolution of 72 × 72 dpi using a digital camera, and the morphology of the mesenteric vein was analyzed using PhotoShop 5.5 (Adobe) image software. To evaluate the total length of the veins, all veins were traced in the region of interest with a one-pixel width line on a computer display and the number of pixels was calculated using histograms. To examine the ratio of vascular length to intestinal length, the length of intestine was also expressed in the number of pixels. After the mesenteric vessels had been analyzed, the jejunum was straightened to measure its length. Then it underwent microscopic observation with hematoxylin-eosin staining. Results were expressed as mean ± SEM. Dunn's test or Wilcoxon's test was performed, and statistical significance was set at P < 0.05.
Perfusion fixation, lectin binding, and peroxidase staining of the jejunum were performed according to previously described methods (Thurston et al., 1998). After peroxidase staining, the jejunum was dehydrated with alcohol and embedded in glycerol.
Whole mount immunohistochemistry.
Wild-type and transgenic embryos were isolated in PBS, fixed in 4% paraformaldehyde in PBS, and stored in 100% methanol at −20°C. The embryos were bleached, rehydrated, and incubated with 2% skim milk and 0.1% Triton X-100 in PBS. Finally, the embryos were incubated with a rat antibody against PECAM-1 (Pharmingen) diluted by 1:100 and a color reaction was performed using the LSAB kit (Dako).
Organs were initially fixed in 4% paraformaldehyde and then in 2.5% glutaraldehyde in cacodylate buffer, pH 7.4, treated with 1% cacodylate-buffered (0.1 M) OsO4, embedded with 1% Epon812, and sectioned. Finally, ultrathin sections were stained with saturated uranyl acetate and lead citrate and examined by electron microscopy.
In situ hybridization.
Slides for in situ hybridization were generated by the frozen section according to established methodology. The sequence of the 18-mer PNA oligomer used as a probe was 5′-GTTATCAACATTTCCACG-3′. A randomly generated pool of PNA oligomers of the same length (18nt) was used as a negative control probe. The GenPoint kit (Dako) was used for in situ hybridization and subsequent color reaction. Following the color reaction, the slides were counterstained with Light Green.
HUVECs were purchased from Toyobo and cultured according to the manufacturer's protocol. For SDF-1 administration, cells were confluently cultured and supplied with 50 ng/mL of murine SDF-1 (Toyobo). Then the cells were harvested every 6 hr for mRNA preparation with TRIzol (Sigma).
The authors thank Jun-ichi Miyazaki for the gift of the transgenic vector, T. Jike for technical assistance, A. Yamashita and T. Yamashita for technical advice, and K. Hirayanagi for advice and assistance with statistical analysis.