Gene deletion studies in mice have led to the identification of numerous genes that are critical to embryonic blood vessel formation. In virtually all of these studies, embryonic vascular defects have been attributed to either aberrant angiogenesis, failed remodeling of primary vascular networks, or impaired mural cell investment. Only three genes, VEGFR-2/Flk-1, fibronectin, and cytochrome P450 reductase/Cpr, have emerged from such studies as being critical to vasculogenesis (Fong et al., 1995; Shalaby et al., 1995; Yang et al., 2000), the process by which endothelial progenitors (angioblasts) assemble and form endothelial tubes. Undoubtedly, there are more genes that are critical to the process of vasculogenesis.
There are many reasons why targeted gene deletion has not led to identification of more vasculogenesis genes. A major reason is that vasculogenesis is generally assessed in mutant vertebrate embryos at a single stage despite the fact that the process of vasculogenesis is initiated in the embryo at different places and times. The first blood vessels of the mouse begin to form in the yolk sac at 6–6.5 dpc (Drake and Fleming, 2000). Later in development (i.e., 7–7.5 dpc), vasculogenesis is initiated within the embryo proper, with blood vessels appearing in the following order: endocardium, primary vascular networks lateral to the midline, paired dorsal aortae, head and cardinal vessels. It is of central importance to understand that nascent blood vessels in one region of a mutant embryo can appear abnormal while those in regions that were initiated at a later stage are normal. This is exemplified by embryos deficient in VE-cadherin, in which blood vessels of the yolk sac are abnormal at 8.5 dpc, whereas the dorsal aortae at this stage appear normal (Crosby et al., 2005). However, by 9.5 dpc, the dorsal aortae of VE-cadherin nulls are morphologically abnormal as well. Accordingly, assessment of the role of a given gene in vasculogenesis can benefit from a systematic evaluation of the spatial and temporal effects of gene ablation on embryonic blood vessel formation.
Here we describe a method for scoring a gene's importance to vasculogenesis based on spatiotemporal vascular defects occurring in response to its inactivation. Using this approach, we retrospectively evaluated embryonic vascular anomalies in numerous mouse knockout studies and ranked the targeted genes according to their importance in vasculogenesis. Based on this analysis, a novel set of genes was identified as being critical to vasculogenesis.
MATERIALS AND METHODS
Approach for Ranking of Embryonic Vascular Anomalies (AREVA)
We have developed a method for assessing the importance of a gene to the process of vasculogenesis based on phenotypic analysis of embryos deficient in the expression of the given gene. The method is referred to as the approach for ranking of embryonic vascular anomalies (AREVA) and is designed to tabulate a numeric score for a gene based on the developmental stage at which mutant embryos display vascular defects at three specific sites of vasculogenesis. The three embryonic sites include the yolk sac, endocardium, and paired dorsal aortae. At each site, blood vessel formation is assessed at three stages of development (i.e., 6–6.5, 7–7.5, and 8–8.5 for yolk sac vessels and 7–7.5, 8–8.5, and 9–9.5 dpc for the endocardium and dorsal aortae). The ranges of development were chosen as they represent periods in which angioblasts assemble to form nascent blood vessels within each site (Drake and Fleming, 2000; Crosby et al., 2005). The endpoints for each period represent the stage at which angioblast populations in each vasculogenic region are no longer present, having been converted into endothelial cells (Drake and Fleming, 2000).
Measures of vascular abnormality include gross morphological anomalies such as absence, enlargement, or reduced diameter of blood vessels and failure of capillary-like networks to be remodeled into larger-caliber blood vessels. For assessment of the earliest events in vasculogenesis, morphological aspects of the process including angioblast aggregation and primary capillary-like network formation need to be evaluated. One means to accomplish this is to perform immunofluorescent microscopic examination using antibodies that detect angioblasts and early endothelial cells. For example, angioblasts can be detected as SCL/Tal-1 and Flk-1 double positive cells; early endothelial cells can be detected as either Flk-1 and PECAM-1 double positive cells or SCL/Tal-1 and PECAM-1 double positive as previously described (Drake and Fleming, 2000; Argraves et al., 2002). A template of procedures for examination of the early-stage mouse vasculature, including embryo dissection and whole-mount immunolabeling, has been described elsewhere (Drake and Fleming, 2000; Crosby et al., 2005).
The AREVA scoring algorithm is weighted such that vascular defects at each of the sites receive positive scores, whereas normal blood vessels at given sites receive negative scores. The scores for vascular defects are further weighted such that early stage defects receive higher scores than later stage defects. For example, at 6–6.5 dpc, abnormal vessels in the yolk sac receive a 3 and normal vessels receive a −3. At 7–7.5 dpc, abnormal vessels in the yolk sac receive a 2 and normal vessels receive a −2. At 8–8.5 dpc, abnormal vessels in the yolk sac receive a 1 and normal vessels receive a −1. Since the endocardium and dorsal aortae initiate formation later than the yolk sac vessels, their scoring period ranges from 7 to 9.5 dpc. At 7–7.5 dpc, abnormal vessels in the endocardium and dorsal aortae receive a 3 and normal vessels receive a −3. At 8–8.5 dpc, abnormal vessels in the endocardium and dorsal aortae receive a 2 and normal vessels receive a −2. At 9–9.5 dpc, abnormal vessels in the endocardium and dorsal aortae receive a 1 and normal vessels receive a −1. Given that this is a retrospective analysis, some sites of vascular development and/or developmental stages are not reported and therefore cannot receive a normal or abnormal score. Under these circumstances, the stages receive a score of 0.
By evaluating blood vessel morphology at each of the three sites and assigning scores for each developmental stage, an average score can be attained for each site and an overall average score obtained for all three sites combined. Based on average combined scores, genes can be ranked. An overall average score of zero is the boundary between genes that are considered important in vasculogenesis (those scoring > 0) and genes likely not having critical roles in vasculogenesis (those scoring ≤ 0).
Compilation of Genes Associated With Vascular Defects
We have reviewed papers describing gene knockouts that produce embryonic lethality attributable to vascular defects. As a result, a set of 79 single gene deletions and 5 double gene deletions were identified (Table 1). Through compilation of embryonic vascular phenotypes described in these studies, it became apparent that yolk sac blood vessels, dorsal aorta, and endocardium were most often evaluated. Of the 100 knockout studies listed in Table 1, 28 reported vascular phenotypes in each of the three sites and 11 reported phenotypes in only two of the three sites.
Table 1. Compilation of genes that when deleted cause embryonic lethality with associated vascular defects
1The genes listed in this table are ordered according to the stage in development at which lethality occurs in embryos deficient in the expression of the indicated gene. Several genes (i.e., ERK5, Tissue Factor, Notch 1 and Endoglin) are each listed more than once in the table due to the fact that stage of lethality differs between separate reports. When two genes are listed together, (i.e., Neuropilin/ Nrp 1&2, Hey 1&2, Notch 1&4, Presenilin 1&2 and ld 1&3) the phenotype indicated corresponds to that of the double knockout. (+/−), indicates the phenotype described was in embryos heterozygous for the targeted gene deletion.
Meta-Analysis of Knockout Studies Using AREVA
AREVA analysis was performed on those gene knockout studies in which embryonic vascular phenotypes were reported in at least two of the three key sites. As a result, the vascular phenotype of embryos from 12 gene deletion studies attained an average score greater than 0 (Table 2). The genes in these studies can be considered as having potentially critical roles to the process of vasculogenesis. The genes assigned to this category included (listed in rank order from highest to lowest AREVA score): fibronectin, VEGFR-1/Flt-1, VEGFR-2/Flk-1, alpha 5 integrin, Tek/Tie2, VE-cadherin, VEGFA, connexin 45, ShcA, cytochrome P450 reductase, CD148/DEP-1, and EphrinB2 (Table 2).
Table 2. AREVA evaluation of embryonic vascular anomalies resulting from targeted deletion of murine genes
1Of the 100 knockout studies listed in the previous table, Table I, only those in which information was reported for at least two of the three key sites of vasculogenesis were subjected to AREVA analysis and listed in this Table.
2When the status of blood vessels at a given sites/stage was not reported in the knockout study, a score of zero was assigned.
Red shading indicates that there was a report of abnormality at the indicated site and stage.
Blue shading indicates that the vasculature was a reported to be normal at the indicated site and stage. Yellow shading highlights those genes whose AREVA scores are greater than zero, thus defining them as critical to vasculogenesis. (+/−), indicates the phenotype described was in embryos heterozygous for the targeted gene deletion. Apostrophes indicate that more than one report of the targeted deletion of the gene exists and that vascular phenotypes are different.
The relative rank of individual genes in the group of 12 can be only tentatively assigned since there is insufficient information in most of the knockout studies. For example, the status of blood vessels in embryos earlier than 8.0 dpc was not reported for any of the genes in this category. Furthermore, for one of the knockout studies in this group, connexin 45, embryonic vascular defects were only reported in two of the three sites of vasculogenesis. It is also important to note that in addition to the 12 genes whose AREVA score placed them in the category of genes being critical to vasculogenesis, there are several genes whose AREVA scores might increase if additional analysis of knockout mice were performed, particularly those gene knockouts that lead to lethality during periods of peak vasculogenesis (i.e., 8.5–9.5 dpc). Included in this group are single gene knockouts for Tal1/SCL and Notch 1 as well as double gene knockouts for presenilin-1 and -2 and neuropilin-1 and -2. While the early embryonic lethality that occurs in these knockouts may not have a vascular origin, it is of interest that the genes are related to the VEGF-signaling pathway. For example, Tal1/SCL expression is promoted by VEGF (Giles et al., 2005) and neuropilin-1 and -2 are both VEGF receptors. Furthermore, Notch 1 and presenilin-1 and -2 are key components of Notch signaling, which downregulates expression of VEGFR2/Flk-1 (Taylor, 2002).
Many of the Genes Scored as Critical for Vasculogenesis Are Implicated in VEGF Signaling
Upon examination of genes assigned as having critical roles in vasculogenesis, it is evident that the majority of the genes (10 of 12) are related to the VEGF signaling pathway. These genes include fibronectin, VEGFR-1/Flt-1, VEGFR-2/Flk-1, alpha 5 integrin subunit, VE-cadherin, VEGFA, cytochrome P450 reductase, ShcA, CD148/DEP-1, and EphrinB2 (Table 2). For the most part, genes in this group can be categorized as either positive or negative regulators of the VEGF signaling pathway. For example, the genes Flt-1, CD148/DEP-1, VE-cadherin, and EphrinB2 are negative regulators in that Flt-1 is a decoy receptor for VEGF, CD148/DEP-1 suppresses VEGF signaling by dephosphorylating Flk-1 (Grazia Lampugnani et al., 2003), VE-cadherin inhibits VEGFR-2/Flk-1 phosphorylation (Grazia Lampugnani et al., 2003), and EphrinB2 inhibits VEGF-induced Ras/mitogen-activated protein kinase activation (Kim et al., 2002).
By contrast, fibronectin, alpha 5 integrin subunit, Flk-1, VEGFA, cytochrome P450 reductase, and ShcA gene products have been implicated as being positive VEGF regulators. For example, fibronectin has been shown to promote VEGF-induced differentiation of peripheral blood-derived endothelial progenitors to endothelial cells (Wijelath et al., 2004). Fibronectin expression is also augmented in response to VEGF signaling (Kazi et al., 2004). Since the alpha 5 integrin is a subunit of the fibronectin-binding receptor, α5β1, it is reasonable to conclude that alpha 5 integrin, like fibronectin, is also a positive regulator of VEGF signaling. Cytochrome P450 reductase promotes the expression of both VEGF and Flk-1, possibly via its regulatory effects on retinol/retinoic acid metabolism and consequences on RXR/RAR transcriptional activation (Otto et al., 2003). The adaptor protein ShcA, a substrate for many tyrosine kinases, including VEGFR-3/Flt4 (Fournier et al., 1999), is phosphorylated in response to VEGF stimuli (Kroll and Waltenberger, 1997). Furthermore, ShcA physically associates with VE-cadherin in a VEGF-dependent manner (Zanetti et al., 2002). The significance of the ShcA-VE-cadherin association remains to be established, but it may result in suppression of ShcA phosphorylation leading to reduced VEGF signal transmission through the MAP kinase pathway.
Targeted Deletion of Some Vasculogenesis-Critical Genes Results in an Augmentation of VEGF Expression and Formation of Sinusoidal Blood Vessels
Of the 12 genes assigned to be important to vasculogenesis, 8 genes (fibronectin, VEGFR-1/Flt-1, connexin 45, VEGFA, Tek/Tie2, ShcA, CD148/DEP-1, and EphrinB2) when deleted lead to the formation of sinusoidal blood vessels in the yolk sac (Dumont et al., 1994; Sato et al., 1995; Carmeliet et al., 1996a; Ferrara et al., 1996; George et al., 1997; Adams et al., 1999; Fong et al., 1999; Kruger et al., 2000; Lai and Pawson, 2000; Gerety and Anderson, 2002; Takahashi et al., 2003; Gale et al., 2004; Krebs et al., 2004). Such anomalies are consistent with VEGF-mediated vascular hyperfusion, a phenomenon in which capillary-sized vessels are replaced by larger sinusoidal vessels in response to augmentation of VEGF levels either as a result of exogenous VEGF administration (Drake and Little, 1995; Argraves et al., 2002), transgenic VEGF overexpression (Benjamin and Keshet, 1997; Zeng et al., 1998; Dor et al., 2002; Eremina et al., 2003), or in response to hypoxia (Kotch et al., 1999). Of the 8 genes that displayed vascular hyperfusion when deleted, VEGF levels were only evaluated in Flt-1 and CD148/DEP-1 deletion mutants (Fong et al., 1999; Takahashi et al., 2003). In each of these mutants, VEGF levels were significantly increased. Increased VEGF expression is an expected outcome in Flt-1 and CD148/DEP-1 nulls considering that VEGF transcription is promoted by VEGF (Vega-Diaz et al., 2001) and both proteins are negative regulators of VEGF receptor signaling. It remains to be established whether any of the remaining knockouts that exhibit hyperfusion also have augmented levels of VEGF.
Targeted Deletion of Candidate Vasculogenesis Genes Results in Cardiac Abnormalities
Eleven of the 12 genes (fibronectin, Flt-1, Flk1, connexin 45, VE-cadherin, VEGF-A, Tek/TIE-2, cytochrome P450 reductase, ShcA, CD148/DEP-1, and EphrinB2), assigned to be important to vasculogenesis, when deleted lead to defects of the endocardium (Table 2). Defective development of the endocardium can be expected to lead to impaired cardiac function, which in turn may have consequences to extracardiac vascular development. For example, impaired cardiac function could result in insufficient oxygenation (hypoxia) of extracardiac embryonic tissues leading to increased VEGF expression (Maltepe and Simon, 1998). Among the 11 gene knockouts displaying endocardial defects, a subset may exhibit extracardiac anomalies in vasculogenesis attributable to hypoxia-induced VEGF expression. However, since the mouse heart begins to contract at 8.5 dpc, cardiac insufficiency cannot influence vasculogenesis that has preceded this stage in sites such as the yolk sac and dorsal aortae, which are formed by 8.5 dpc. Thus, in knockouts of fibronectin, Flt-1, Flk1, connexin 45, and VE-cadherin genes, in which cardiac defects are evident between 8 and 8.5 dpc (George et al., 1993; Carmeliet et al., 1999; Gory-Faure et al., 1999; Kruger et al., 2000; Crosby et al., 2005), abnormalities in the yolk sac vessels that are apparent at this same period were not the result of cardiac insufficiency.
ShcA/Ras/Raf/Mek/Erk Pathway Integrates VEGF Signaling With Additional Signaling Cascades
When the 12 vasculogenesis-critical genes are considered with respect to their involvement in canonical signaling pathways, an integrated network centering on the ShcA/Ras/Raf/Mek/Erk pathway becomes apparent (Fig. 1). Simply described, the integrated network has at its hub the ShcA/Ras/Raf/Mek/Erk pathway with links to fibronectin-α5β1 integrin, angiopoietin-Tie2, EphrinB2/Eph, and Connexin 45 signaling pathways. The adaptor protein ShcA, which promotes Grb2/Sos/Ras-dependent Erk activation, appears to be a central node in the network given that 7 of the 12 vasculogenesis genes (i.e., fibronectin, VEGFA, VEGFR-2/Flk-1, VEGFR-1/Flt-1, VE-cadherin, integrin alpha 5, and Tek/Tie2) have functional relationships that involve ShcA. For example, ShcA interacts with the fibronectin receptor (α5β1) via the integrin alpha 5 subunit (Wary et al., 1996; Mauro et al., 1999), and fibroblasts from ShcA-deficient mice display defective spreading when plated on fibronectin (Lai and Pawson, 2000). Similarly, ShcA also interacts with the tyrosine kinase receptor, Tie2, and mediates angiopoietin-1-induced chemotaxis and sprouting in endothelial cells (Audero et al., 2004). Finally, ShcA also interacts with VEGFR-2 (Zanetti et al., 2002) and VEGF stimulation results in the phosphorylation of the 46, 52, and 66 kDa isoforms of ShcA and the induction of Shc-Grb2 complex formation (Seetharam et al., 1995; Kroll and Waltenberger, 1997). Additionally, Shc binds to the carboxy-terminal domain of VE-cadherin and this interaction exerts a negative effect on Shc phosphorylation by VEGFR-2/Flk-1 (Zanetti et al., 2002).
Three of the 12 vasculogenesis critical genes, EphrinB2, Tek/Tie2, and connexin 45, can regulate components of the Ras/mitogen-activated protein kinase-signaling pathway down stream from ShcA. The Tek/Tie-2 ligand, angiopoietin-1, can increase both Erk1/2 and p38 phosphorylation in endothelial cells (Kim et al., 2002; Harfouche et al., 2003). By contrast, the EphB ligand, EphrinB2, suppresses the VEGF- and angiopoietin 1-induced Ras/mitogen-activated protein kinase pathway in endothelial cells (Kim et al., 2002). Finally, the gap junction protein connexin 45 is required for optimal activation of signal transduction through Erk (Stains and Civitelli, 2005). Taken together, the bulk of the genes identified as critical to vasculogenesis can be integrated into a network of modulators of the ShcA/Ras/Raf/Mek/Erk cascade, a principal pathway for transmitting VEGF-stimulated signal transduction.
The authors thank Paul A. Fleming, Cynthia K. Gittinger, and Dr. Jeremy Barth (Medical University of South Carolina, Charleston, SC) for their assistance with the preparation of this article.