Abstract
Free full text
Akt1 regulates pathological angiogenesis, vascular maturation and permeability in vivo
Abstract
Akt kinases control essential cellular functions, including proliferation, apoptosis, metabolism and transcription, and have been proposed as promising targets for treatment of angiogenesis-dependent pathologies, such as cancer and ischemic injury. But their precise roles in neovascularization remain elusive. Here we show that Akt1 is the predominant isoform in vascular cells and describe the unexpected consequences of Akt1 knockout on vascular integrity and pathological angiogenesis. Angiogenic responses in three distinct in vivo models were enhanced in Akt1−/− mice; these enhanced responses were associated with impairment of blood vessel maturation and increased vascular permeability. Although impaired vascular maturation in Akt1−/− mice may be attributed to reduced activation of endothelial nitric oxide synthase (eNOS), the major phenotypic changes in vascular permeability and angiogenesis were linked to reduced expression of two endogenous vascular regulators, thrombospondins 1 (TSP-1) and 2 (TSP-2). Re-expression of TSP-1 and TSP-2 in mice transplanted with wild-type bone marrow corrected the angiogenic abnormalities in Akt1−/− mice. These findings establish a crucial role of an Akt-thrombospondin axis in angiogenesis.
The formation of a vasculature by vasculogenesis and angiogenesis is essential for embryonic development1, tissue remodeling in adults and the unrestrained growth of tumors2. Vascular endothelial growth factor (VEGF) is widely accepted as a primary inducer of angiogenesis, but other factors have recently emerged as important regulators, including extracellular matrix proteins and their integrin receptors3. The angiogenic response to VEGF begins with vascular leakage and the deposition of a provisional matrix consisting of blood-derived components; the activation of proteases eventually degrades the extracellular matrix, releasing endothelial cells. The migration and proliferation of endothelium results in the formation of an initial capillary network, which becomes mature upon recruitment of mural cells3. Many of the biological activities ascribed to VEGF involve the activation of the phosphoinositide 3-kinase (PI3 kinase)-Akt signaling cascade. The inter-relationship between VEGF and the Akt pathway in vivo is complex, because PI3 kinase–Akt activation in turn mediates VEGF production in tissues4. Akt is known to regulate short-term cellular processes, its activation occurs almost immediately upon stimulation by several growth factors and activated Akt can then phosphorylate multiple substrates. Notably, Akt controls long-term responses by regulating protein expression at the level of transcription and translation5. Thus, as a result of the complexity of the signaling network downstream of Akt, its role in pathophysiological processes remains elusive.
Recent studies conducted using transgenic and knockout animals have indicated that our understanding of Akt signaling mechanisms is far from complete. Although the phenotype of animals lacking single Akt isoforms is relatively mild6, Akt1-Akt2 double-knockout mice die shortly after birth7. Although reports of vascular functions in Akt-deficient and transgenic mice are limited, it is clear that the role of Akt signaling, though not predictable, is of biological significance8.
To delineate the role of the Akt pathway in the regulation of the formation and integrity of the vasculature, we used mice with a genetic disruption of the Akt1 gene. Because of the numerous downstream targets of Akt and the possibility of distinct and opposing short-term and long-term consequences of Akt activation9, an analysis of the vascular phenotype and its underlying mechanisms represents an experimental challenge. We therefore focused on the tissue responses to VEGF administered through multiple routes, excluding models that involve the induction of endogenous VEGF, to streamline our search for downstream effects of Akt.
RESULTS
Akt1 is the predominant isoform of Akt in endothelial cells
We investigated the expression of all three Akt isoforms in endothelial cells from the aorta, lung and Matrigel implants and found that Akt1 is the predominant isoform in these cells (Fig. 1a). In response to VEGF-A stimulation, active Akt (Ser473-phosphorylated Akt) levels in endothelial cells from Akt1−/− mice were less than 25% of those in endothelial cells from wild-type animals (Fig. 1b). This finding suggests a prominent role for Akt1 in VEGF-induced responses. Western blots for individual Akt isoforms show that the absence of Akt1 resulted in only minor changes in Akt2 and Akt3 expression in various tissues (Supplementary Fig. 1 online). Moreover, in the absence of Akt1, total Akt activity, measured by phosphorylation of GSK-3β, was reduced by at least three- to fourfold in nonstimulated endothelial cells as well as in response to VEGF (Fig. 1c).
Deficiency of Akt1, the predominant Akt isoform in endothelial cells, impairs their function ex vivo. (a) Western blot analyses and band densitometry of total Akt expression (using panAkt antibodies) in endothelial cells isolated from lungs, aorta and Matrigel plugs and in skin of wild-type (filled bars) and Akt1−/− (open bars) mice. (b) Western blot analyses of total phosphorylated Akt in lung endothelial cells from wild-type and Akt1−/− mice in response to 20 ng/ml VEGF. (c) Western blot analyses of phosphorylated GSK-3β in lung endothelial cells from wild-type and Akt1−/− mice in the presence and absence of VEGF (20 ng/ml). (d) Micrographs of representative aortic ring microvessels from wild-type (top) and Akt1−/− (bottom) animals grown in the presence of VEGF. Scale bars, 20 μm. (e) Quantification of vascular outgrowth in the presence or absence of VEGF (30 ng/ml) was performed by FACS analysis using CD31-specific antibodies. MFI, mean of fluorescence intensity. (f,g) Endothelial cell migration toward vitronectin (f) and fibronectin (g) was stimulated by 25 and 100 ng/ml of VEGF as indicated. (h) Permeability of endothelial monolayers in vitro. Wild-type and Akt1−/− endothelial cell monolayers were grown in transwells for at least 1 week to reach confluency and then stimulated with VEGF (30 ng/ml) or were unstimulated. n refers to number of times each experiment was repeated for each condition.
Akt1 in VEGF-induced angiogenic responses in vitro and ex vivo
In a frequently used ex vivo aortic ring assay, the average length and branch number of outgrown microvessels were reduced by at least 40% using rings from Akt1−/− animals compared to those from wild-type animals (Fig. 1d). In an independent approach, we corroborated these observations by quantifying the outgrowth of CD31-expressing endothelial cells using FACS analysis (Fig. 1e). Based on the results of thymidine-incorporation assays, wild-type endothelial cells showed a 3.2-fold increase in proliferation in response to serum. In contrast, Akt1−/− endothelial cells showed only a modest 1.4-fold increase (data not shown). Thus, the proliferation rate of Akt1−/− endothelial cells in vitro is reduced.
As cell migration is an essential step in angiogenic response, we assessed the effect of Akt1 deficiency on endothelial cell migration. VEGF exerted a pronounced effect on the migration of wild-type endothelial cells on various matrix proteins, whereas the migration of Akt1−/− endothelial cells in response to VEGF was significantly impaired (Fig. 1f,g and Supplementary Fig. 1 online). We have previously shown that VEGF induces integrin activation in endothelial cells10, which can be quantified as an increase in fibrinogen binding. In a soluble ligand–binding assay, Akt1−/− endothelial cells bound three times less fibrinogen than their wild-type counterparts (Supplementary Fig. 1 online). We observed no changes in the expression levels of major integrins (αvβ3, αvβ5, α5β1) and VEGF receptors (data not shown). Impaired integrin function often results in decreased extracellular matrix assembly and deposition11, which in turn might affect the integrity of the endothelial monolayer and its permeability, an essential part of the angiogenic response12. As anticipated, Akt1−/− endothelial cell monolayers were ~50% more leaky than their wild-type counterparts in the presence and in the absence of VEGF stimulation (Fig. 1h).
Enhanced pathological angiogenesis in Akt1−/− mice
To investigate the role of Akt1 in neovascularization in vivo, we injected mice subcutaneously with Matrigel with or without VEGF. We observed larger plugs with a threefold increase in hemoglobin content in the presence of VEGF in Akt1−/− mice compared to wild-type mice (Fig. 2a). Quantification of endothelial cells in Matrigel showed that VEGF-induced vascular proliferation in Akt1−/− mice was enhanced by 40% (Fig. 2b), suggesting that the permeability of Akt1−/− blood vessels contributed greatly to the angiogenic response.
Enhanced in vivo angiogenesis in Akt1−/− mice. (a,b) Angiogenesis in Matrigel implants stimulated with VEGF. (a) Hemoglobin content in Matrigel implants from Akt1−/− mice was higher compared to that from wild-type mice. Inset shows the microscopic appearance of 5-d-old Matrigel implants with 60 ng/ml VEGF from wild-type (top) and Akt1−/− (bottom) mice (original magnification, ×2). (b) Increased endothelial cell infiltration into Matrigel in Akt1−/− versus wild-type mice. (c–e) Angiogenesis induced by implanted tumors. (c) Comparison of CD31 staining of tumors grown in wild-type (left) and Akt1−/− mice (right). Scale bars, 20 μm. (d) Image analysis shows the increased density of blood vessels (the number of vessels per tissue area) in tumors grown in Akt1−/− versus wild-type mice. (e) Comparison of vascular area in tumors implanted in wild-type and Akt1−/− mice. (f) A representative photograph of the necrotic area in the center of tumor grown in wild-type mice (left). No necrosis was observed in tumors from Akt1−/− mice (right). Sections were stained with hematoxylin and eosin. Scale bars, 50 μm. (g) Increased number of CD105-positive blood vessels (green fluorescence) in frozen sections of tumors grown in Akt1−/− mice (middle) compared to wild-type mice (left). Right panel shows costaining for CD105 (green) and laminin (red) in the tumor blood vessels in wild-type mice. Nuclei are stained with DAPI. Scale bars, 10 μm. (h) Quantitative analysis of CD105-positive vessels in tumors from Akt1−/− mice compared to wild-type mice. n refers to the number of mice in each group.
Pathological angiogenesis induced by tumor implantation was assessed in wild-type and Akt1−/− mice. In these experiments, B16F10 melanoma cells were implanted subcutaneously. Tumors grown in Akt1−/− mice were characterized by a considerable increase in the number of blood vessels compared to tumors grown in wild-type mice (Fig. 2c). On the periphery of tumors formed in wild-type mice, a substantial vasculature was present, but it was markedly decreased in the center. In contrast, tumors grown in Akt1−/− mice had a relatively equal distribution of blood vessels and the lack of vasculature in the center was not apparent. Notably, tumors grown in Akt1−/− mice had a significantly higher vascular density (Fig. 2d) and vascular area compared to wild-type controls (Fig. 2e). No differences were found in the vascularity of normal skin in wild-type versus Akt1−/− mice (data not shown), suggesting that the enhanced tumor angiogenesis was not the result of preexisting differences in blood vessel density. Notably, in two-thirds of the tumors grown in wild-type animals, we observed a substantial necrotic core (15% of the total tumor area on average; Fig. 2f). But none of the tumors grown in Akt1−/− mice showed any necrosis, indicating that increased tumor vascularization provided a more efficient blood supply in tumors grown in Akt1−/− compared to wild-type mice.
Akt1-deficient vasculature is immature and leaky
The majority of tumor-induced blood vessels were positive for CD105, a marker of newly formed blood vessels (Fig. 2g). The density of ‘young’ blood vessels was 2.3-fold higher in tumors implanted in Akt1−/− versus wild-type mice (Fig. 2h). Blood vessel maturation measured by smooth muscle–cell actin (SMA) staining was considerably reduced in the absence of Akt1 (Fig. 3a). In wild-type mice, 17% of total CD31+ blood vessels stained positively for SMA, whereas only 9.5% of blood vessels formed in Akt1−/− mice were mature blood vessels (Fig. 3b). Moreover, immunostaining for laminin and subsequent confocal microscopic analysis showed a reduction of at least 50% in the thickness of basement membrane of blood vessels formed in Akt1−/− mice compared to those formed in wild-type mice (Fig. 3c,d). Details of the ultrastructure of blood vessels formed in wild-type and Akt1−/− mice are presented in Supplementary Figure 2 online. Based on western blot analysis, laminin content in blood vessels was significantly reduced in tumors grown in Akt1−/− mice (P = 0.031; Supplementary Fig. 2 online). Thus, Akt1−/− blood vessels are characterized by a less developed basement membrane and deficient recruitment of mural cells, which in turn could result in the leaky phenotype of the Akt1−/− vasculature. Indeed, plasma leakage, measured as an area positive for plasma-derived fibrin (Fig. 3e), was almost doubled in tumors implanted in Akt1−/− mice (Fig. 3f).
Neovasculature in Akt1−/− mice is immature and leaky. Immunohistochemistry and image analysis of tumors implanted into wild-type and Akt1−/− mice. (a) Costaining for SMA (green) and CD31 (red) in tumors from wild-type and Akt1−/− mice. Nuclei are stained with DAPI. Arrowheads indicate SMA-positive vascular structures. Scale bars, 50 μm. (b) Quantification of the data presented in a. (c) Laminin-stained blood vessels in tumors grown in wild-type (top) and Akt1−/− mice (bottom). Insets show higher-magnification images. Scale bars, 20 μm. Inset scale bar, 5 μm. Immunohistochemistry (right panels) shows laminin (red), CD105 (green) and nuclei (blue). Scale bars, 10 μm. (d) Thickness of laminin-positive basement membrane in microvessels formed in wild-type and in Akt1−/− mice. (e) Fibrin deposition (brown) in tumors grown in wild-type (top) and Akt1−/− (bottom) mice. Scale bars, 20 μm. (f) Quantification of fibrin-positive area shown in e. (g,h) Angiogenesis induced by VEGF-A165 or GFP (control) adenoviruses. (g) Comparison of vascular area in vWF-stained sections of wild-type and Akt1−/− skin upon stimulation with VEGF-A165. Representative images are in Supplementary Figure 2 online. (h) Comparison of SMA-positive vascular area in wild-type and Akt1−/− skin 7 d after injection of Ad-VEGF-A. Representative images are in Supplementary Fig. 2 online.
We observed similar results in a second model of transient angiogenesis induced by adenovirus carrying the gene that encodes VEGF-A (Ad-VEGF-A)13. In these experiments, Ad-VEGF-A but not control Ad-GFP stimulated a considerable angiogenic response, which reached a maximum 7–10 d after adenovirus injection (data not shown). At this time point, Ad-VEGF-A–induced vascularization in Akt1−/− mice was higher than that in their wild-type counterparts (Fig. 3g) Furthermore, in this model of angiogenesis, the proportion of mature SMA-positive blood vessels was at least 50% less in Akt1−/− mice compared to wild-type counter-parts (Fig. 3h).
Next, we used angiogenic growth factors with different activities to stimulate vascular permeability. VEGF-D, a growth factor that does not stimulate vascular permeability14, caused a ~2-fold increase in vascular area in wild-type mice but not in Akt1−/− mice (Fig. 4a). A truncated form of VEGF-D (i.e., VEGF-DΔNΔC), which is able to induce vascular permeability14, stimulated increases in vascular area of 2.4- and 4.3-fold in wild-type and Akt1−/− mice, respectively (Fig. 4a). These results show that enhanced vascular leakage in Akt1−/− mice contributes greatly to increased angiogenic response.
Characterization of vascular responses in Akt1−/− mice. (a) Comparison of angiogenesis induced by full-length VEGF-D and VEGF-DΔNΔC in wild-type and Akt1−/− mice. (b) Comparison of endothelial cell proliferation in Ad-GFP– or Ad-VEGF-A–treated skin of wild-type and Akt1−/− mice based on Ki-67 staining. Inset shows a representative image of proliferating endothelial cells (arrowheads) in a skin blood vessel. Scale bar, 10 μm. (c) Plasma leakage in ear skin assessed by Evans blue extravasation after treatment with mustard oil, with no difference in control (unstimulated) ears. (d) Vascular permeability in wild-type and Akt1−/− mice in response to Ad-VEGF-A treatment. Fibrin staining shows increased leakage of fibrin in Akt1−/− compared to wild-type mice. Scale bars, 50 μm. (e) Area of fibrin deposition in skins from wild-type and Akt1−/− mice shown in d. (f) Comparison of ear microvessel morphology and leakage sites in ricin-stained whole mounts of ear skin in wild-type and Akt1−/− mice upon stimulation with mustard oil. Dark patches (shown by arrows) indicate sites of exposed basement membrane. There was an increased number of dark patches in blood vessels in Akt1−/− (lower panel, inset) compared to wild-type mice (upper panel, inset). Scale bars, 50 μm. Inset scale bar, 20 μm. (g) Quantitative analysis of the leakage area of wild-type and Akt1−/− blood vessels shown in f. n refers to the number of mice in each group.
As anticipated from previously published studies and from our own in vitro analysis, the basal proliferation rate of endothelial cells in the untreated skin of Akt1−/− mice is much lower than that in wild-type mice, as judged by staining for the proliferation marker Ki67 (Fig. 4b). But 7 d after treatment with Ad-VEGF-A, endothelial cell proliferation increased 2.5- and 15-fold in skin in wild-type and Akt1−/− animals, respectively. Thus, it seems that VEGF-A–stimulated proliferation of endothelial cells in vivo was not impaired by the lack of Akt1.
Notably, the permeability of preexisting blood vessels in response to a proinflammatory stimulus was 2.4-fold higher in Akt1−/− mice compared to wild-type mice (Fig. 4c). In addition, we observed increased deposition of a plasma-derived provisional matrix (which stained positively for fibrin) in Akt1−/− mice in response to Ad-VEGF-A but not to control Ad-GFP (Fig. 4d,e). This finding was corroborated by the results of ricin staining, which visualizes microvessel morphology and the sites of vascular leakage in whole mounts of ear skin15. Upon stimulation, we observed more intensive staining and increased density of leakage sites in the vasculature of Akt1−/− mice than in wild-type mice (Fig. 4f). Quantitative analysis showed increased areas of exposed basement membrane in vessels in Akt1−/− compared to wild-type mice (Fig. 4g). Thus, newly formed and preexisting blood vessels in the skin of Akt1−/− mice are defective and more permeable than those in wild-type mice. In our models, we observed no significant differences in the expression of angiopoietin-1 and angiopoietin-2, major regulators of vascular permeability, between wild-type and Akt1−/− mice (Supplementary Fig. 2 online).
Akt1 deficiency results in changes of extracellular matrix
The observed changes in basement membrane matrix composition prompted us to perform an analysis of several other matrix proteins that are known to be involved in the regulation of angiogenesis. We observed a substantially reduced density of collagen fibrils and a reduced amount of total collagen in the skin of Akt1−/− compared to wild-type mice (Supplementary Fig. 3 online). As a consequence, tumors grown in wild-type but not in Akt1−/− mice were densely packed and showed signs of solid (mechanical) stress and compression generated by interaction of the expanding tumor with the extracellular matrix (Supplementary Fig. 3 online). Thus, the reduced stiffness of the surrounding extracellular matrix, together with increased vascular permeability, may facilitate the growth of microvessels in Akt1−/− mice.
Several features of Akt1−/− mice (abnormalities in collagen fibril organization, increased vascular permeability and enhanced angiogenesis) closely resemble the phenotype of mice deficient in the matricellular proteins thrombospondin 1 (TSP-1) and thrombospondin 2 (TSP-2)16. Accordingly, we performed an analysis of TSP levels. TSP-1 expression in implanted tumors was limited to stromal cells, and was not present in tumor cells (Fig. 5a). The TSP-1–positive area in tumors grown in Akt1−/− mice was significantly smaller than that in wild-type tumors (Fig. 5b). Notably, the tumor vascular density correlated inversely with the expression level of TSP-1 (Fig. 5c and Supplementary Fig. 4 online), suggesting a role for TSP-1 in the regulation of tumor angiogenesis. Western blot analysis of tissue lysates showed reduced expression of both TSP-1 and TSP-2 in tumors grown in Akt1−/− mice as compared to those grown in wild-type mice, as well as reduced expression in the skin of Akt1−/− compared to wild-type mice (Fig. 5d). Endothelial cells isolated from Akt1−/− mice retained their in vivo characteristics and expressed reduced levels of TSP-1 and TSP-2 when compared to endothelial cells from wild-type mice (Fig. 5e and Supplementary Fig. 4 online).
Akt1 deficiency results in reduction of TSP levels. (a) TSP-1 staining in tumors from wild-type (left) and Akt1−/− mice (middle). Right panel shows costaining for TSP-1 (green) and CD31 (red) in tumors from Akt1−/− mice. Scale bar, 20 μm. (b) Comparison of TSP-1 staining in tumors from wild-type and Akt1−/− mice. (c) A reciprocal correlation existed between vascularity and TSP-1 expression in tumors. (d) Western blot and subsequent densitometry shows reduced levels of TSP-1 and TSP-2 in tissues from Akt1−/−compared to wild-type mice. (e) Levels of TSP-1, phosphorylated Akt, Akt1 and β-actin in wild-type and Akt1−/− endothelial cells upon treatment with Ad-myrAkt1 or Ad-GFP determined by western blot. There were increased levels of TSP-1 upon re-expression of Akt1. (f) Luciferase activity measured in HUVECs cotransfected with a Tsp 2 promoter/luciferase construct and retroviral constructs encoding DNAKt1, myrAkt1 or control. (g) Permeability of endothelial cell monolayers from wild-type and Akt1−/− mice in the presence or absence of TSP-1. Exogenous TSP-1 (1 μg/ml) decreased the permeability of Akt1−/− endothelial cell monolayers, whereas it slightly increased the permeability of wild-type endothelial cell monolayers. (h) The consequences of downregulation of TSP-1 by siRNA on the permeability of endothelial cell monolayers. Endothelial cells were treated with TSP-1 siRNA lentivirus (siTSP-1) or luciferase siRNA (siLuc) as a control. Inset shows TSP-1 levels after siRNA treatment. The permeability of endothelial cells was stimulated by 30 ng/ml VEGF-A165 as indicated. n refers to number of times each experiment was repeated for each condition.
Akt1 controls expression of TSPs
Decreased levels of TSP-1 in Akt1−/− mice correlated with the extent of Akt activation (Fig. 5e). Upregulation of Akt activity in Akt1−/− endothelial cells by treatment with adenovirus encoding a constitutively active myristolated (myr) form of Akt1 (Ad-myrAkt1) resulted in a considerable increase in TSP-1 expression (Fig. 5e). Next, we transfected endothelial cells with a plasmid in which the Thbs2 (encoding TSP-2) promoter (starting at nucleotide −2,701) controls expression of a luciferase reporter gene and cotransfected with dominant negative and constitutively active forms of Akt1. Inhibition of Akt1 activity by contransfection with retroviral constructs encoding a dominant negative Akt (DNAkt1) resulted in a nearly threefold decrease in luciferase levels (Fig. 5f). In contrast, the expression of myrAkt1 produced a modest but significant increase in luciferase levels over control, showing that Akt1 directly regulates expression of TSPs in endothelial cells.
Notably, exogenous TSP-1 was able to modulate the permeability of an endothelial cell monolayer (Fig. 5g). Moreover, downregulation of endogenous TSP-1 levels by siRNA in wild-type endothelial cells caused an increase in endothelial cell permeability (Fig. 5h), providing direct evidence that TSP-1 is a crucial factor in vascular permeability.
Role of an Akt1-TSP axis in angiogenesis
To confirm a direct role of Akt signaling in the regulation of endothelial cell permeability, we transfected Akt1−/− endothelial cells with Ad-myrAkt1 or an adenovirus carrying the gene encoding green fluorescent protein (GFP; Ad-GFP, control). Restoration of Akt1 activity with myrAkt1 in Akt1−/− endothelial cells reversed the leaky phenotype of these cells, whereas treatment with Ad-GFP had no effect (Fig. 6a). Notably, the hyperpermeability of blood vessels in Akt1−/− mice was rescued in vivo by prolonged treatment with Ad-myrAkt1 (Fig. 6b). Together, these data indicate that Akt1 signaling regulates vascular permeability and maturation, which in turn influence angiogenic responses in vivo.
An Akt1-TSP axis regulates vascular permeability and angiogenesis. (a) Comparison of the endothelial cell permeability of wild-type and Akt1−/− endothelial cells after treatment with Ad-myrAkt1 or Ad-GFP. Restoration of Akt resulted in the decrease of endothelial cell permeability to the level observed in wild-type endothelial cells. (b) Effect of prolonged Ad-myrAkt1 treatment on vascular permeability in Akt1−/− mice. Vascular leakage of Evans blue was stimulated by mustard oil (n = 6 mice per group). (c–h) TSP re-expression corrects the angiogenic phenotype of Akt1−/− mice. Primary endothelial cells and fibroblasts transfected with TSP-1 and TSP-2 were subcutaneously injected into Akt1−/− chimeric mice reconstituted with wild-type bone marrow followed by implantation of a mixture of tumor cells and TSP-overexpressing cells (Supplementary Methods online). (c) Western blot analysis shows increased expression of TSP-1 and TSP-2 in vitro in primary endothelial cells and fibroblasts cotransfected with plasmids carrying genes encoding TSP-1 and TSP-2. Increased TSP-1 and TSP-2 expression was also seen in the tumors grown in vivo. (d) Increased levels of TSP-1 (green staining, upper panels) and a decreased number of CD31-stained blood vessels (red, lower panels) in tumors upon re-expression of TSP-1 and TSP-2. Scale bar, 20 μm. (e) Effect of TSP-1 and TSP-2 on tumor weight. (f) Area of necrosis in tumors upon re-expression of TSP-1 and TSP-2. (g) Vascular density upon expression of TSP-1 and TSP-2 in tumors shown in d. (h) Expression of TSP-1 and TSP-2 resulted in decreased endothelial cell proliferation in vivo based on Ki67 staining (Fig. 4).
In further analysis of the role that TSPs might have in abnormal angiogenesis in Akt1−/− mice, we considered that two major sources of TSPs could influence the course of an angiogenic response: TSP-1 released from platelets, and TSP-1 and TSP-2 expressed by endothelial and stromal cells17. Accordingly, to examine the contribution of platelets18 and to assess the role of bone-marrow derived components in angiogenesis, we replaced the bone marrow of Akt1−/− mice with wild-type cells. Although this procedure diminished tumor angiogenesis and tumor cell survival somewhat in Akt1−/− mice, it did not result in a complete rescue of the phenotype. The vascular area in tumors quantified by CD31 staining, and tumor weight were decreased by 20% and 17%, respectively, in chimeras compared to Akt1−/− mice (data not shown). Thus, although bone marrow–derived cells are involved in the highly angiogenic phenotype of Akt1−/− mice, the local tissue environment seems to have a much more prominent role. In a further attempt to correct the phenotype of Akt1−/− mice, we restored TSP levels by injecting cells expressing TSP-1 and TSP-2. Primary syngeneic endothelial cells were transfected with murine TSP-1 and TSP-2 and injected subcutaneously into chimeric Akt1−/− mice transplanted with wild-type bone marrow (Supplementary Methods online). Two weeks later, a mixture of melanoma cells and TSP-overexpressing cells were inoculated into the same areas, and tumors were collected after 1 week. Western blot analysis showed an increase in levels of both TSP-1 and TSP-2 in endothelial cells in vitro, as well as in tumors grown in vivo (Fig. 6c), and this result was further confirmed by immunostaining of tumor sections (Fig. 6d). As anticipated, increased TSP expression significantly suppressed tumor growth and tumor cell survival, as judged by tumor weight (Fig. 6e) and the area of necrosis in vivo (Fig. 6f). Notably, the density of blood vessels, as well as the proliferative rate of endothelial cells in vivo, was significantly reduced by TSP expression (Fig. 6d,g,h). Taking these data together, it seems that local re-expression of TSP-1 and TSP-2, combined with transplantation with wild-type bone marrow, is able to correct the phenotype of Akt1−/− mice.
Role of an Akt1-eNOS pathway in angiogenesis
Endothelial nitric oxide synthase (eNOS) is a known downstream target of Akt and might also contribute to the angiogenic phenotype of Akt1−/− mice, as described in a recent study19. Indeed, we observed a modest decrease in the basal levels of active eNOS in endothelial cells isolated from Akt1−/− mice when compared to wild-type mice. But the presence of compensatory mechanisms is evident, as we observed only a small difference upon stimulation with VEGF (Supplementary Fig. 5 online). To examine the role of eNOS (encoded by the Nos3 gene), we assessed tumor-induced angiogenesis in Nos3−/− mice (Supplementary Figs. 5 and 6 online). Similar to Akt1−/− mice, tumor blood vessels in Nos3−/− mice were considerably less mature than in wild-type, with an apparent lack of pericyte recruitment as evidenced by tissue costaining for CD31 and SMA (Supplementary Fig. 6 online). But, in contrast to Akt1−/− mice, the vascular area in tumors grown in Nos3−/− mice was similar to that in wild-type mice (Supplementary Fig. 6 online). Notably, we did not observe any changes in the expression of TSPs in Nos3−/− mice (data not shown). Thus, the lack of pericyte recruitment and vascular maturation observed in newly formed blood vessels of Akt1−/− mice might result from a lack of eNOS activity in Akt1−/− endothelial cells (Supplementary Figs. 5 and 6 online). But enhanced angiogenesis and tumor survival in our models seem to be regulated by the composition of the extracellular matrix.
DISCUSSION
The Akt pathway might regulate angiogenic responses by several distinct and possibly counterbalancing mechanisms. For example, Akt is known to mediate hypoxia-induced expression of VEGF in vitro and in vivo20,21, but its long-term activation decreases VEGF expression and vascularity in cardiac tissues9. Here, we find that a number of endothelial cell responses involved in angiogenesis, including capillary formation ex vivo, endothelial cell proliferation and endothelial cell migration in response to VEGF in vitro, were impaired in Akt1−/− endothelial cells, but that the permeability of an endothelial monolayer was increased, which would be expected to have a proangiogenic effect. Although these data again point to the multiple and complex roles of Akt1 in endothelial cell responses induced by VEGF, they raise the question of which effects of Akt1 will determine the course of the angiogenic response in vivo. To address this question, we assessed three distinct in vivo models of angiogenesis—angiogenesis in Matrigel, in implanted tumors and upon stimulation with Ad-VEGF. Our results showed that the lack of Akt1 causes enhanced angiogenesis. Thus, we conclude that the overall effect of Akt1 on angiogenesis is suppressive. In contrast, a recent study19 reported impaired angiogenesis in Akt1−/− mice in hindlimb ischemia as well as upon treatment with Ad-VEGF-A. Because of the transient nature of Ad-VEGF-A–induced angiogenesis13, it is possible that initial neovascularization (5 d after treatment)19 might be delayed in Akt1−/− as compared to wild-type mice. In this study, we showed that at the peak of the angiogenic response, which occurs 7–10 d after Ad-VEGF-A treatment, Akt1−/− mice developed extensive vasculature. To reconcile these results, we propose that Akt1 might have different roles at different stages of angiogenesis. Moreover, the short-term and long-term consequences of Akt1 activation might be distinct. In support of this interpretation, another recent study showed that chronic but not short-term overexpression of Akt1 in cardiac tissues suppresses angiogenesis9, a finding consistent with our results.
In seeking to define the mechanisms of the effects of Akt1 on angiogenesis, we found that neovasculature in Akt1−/− mice was defective in pericyte recruitment, resulting in reduced maturation and a significant decrease in laminin deposition and basement membrane thickness. This feature of blood vessels in Akt1−/− mice might be a result of the insufficient short-term activation of one of the downstream targets of Akt, eNOS, as a similar defect in maturation was observed in eNOS-deficient (Nos3−/−) mice22. We observed deficient maturation and increased permeability of blood vessels in Akt1−/− mice in all models of angiogenesis studied here. The direct role of Akt1 in the maintenance of vascular permeability was further supported by our observation that prolonged but not short-term expression of Akt1 corrected excessive vascular permeability in vitro and in vivo.
The majority of angiogenic growth factors, including VEGF-A, initiate their action by stimulation of vascular permeability, resulting in the deposition of a ‘provisional’ matrix for endothelial cell migration and formation of endothelial tubes23. An increased amount of provisional matrix might promote endothelial cell migration and stimulate a more profound angiogenic response. Notably, only growth factors that induce vascular leakage were able to generate a highly angiogenic phenotype in Akt1−/− mice. It has also been reported that bone marrow-derived cells might contribute to angiogenesis24. The replacement of bone marrow of Akt1−/− mice with wild-type cells confirmed the role of bone marrow–derived components in our model, but we did not observe complete rescue.
Another factor that seems to contribute to the increased angiogenesis and increased tumor cell viability in Akt1−/− mice is a reduction in skin collagen content. High collagen density increases the mechanical resistance of the matrix, which in turn reduces tumor growth; conversely, a loose matrix permits the formation of large, well-vascularized tumors25,26. Abnormally loose extracellular matrix represents an easier target for proteases, which have an essential role in the regulation of angiogenesis27. The same consideration applies to the laminin-deficient basement membranes of blood vessels in Akt1−/− mice, which might permit the accelerated release and mobilization of endothelial cells from preexisting vascular structures.
At a mechanistic level, vascular leakage and enhanced angiogenesis in Akt1−/− mice are directly linked to abnormalities in extracellular matrix composition, which, in turn, are long-term consequences of deficient Akt1 activation, and, therefore, might not be observed in models that involve short-term modulation of the Akt1 pathway. The regulation of extracellular matrix content by Akt1 might provide an explanation for the recently reported cardiac dysfunction and hyper-trophy in mice that overexpress Akt19. An essential component of this biologically important connection is an observed decrease in the amount of TSP-1 and TSP-2 in the tissues of Akt1−/− mice. Indeed, our results showed that Akt1 directly controls the expression of TSPs at the transcriptional level. Regulation of extracellular matrix seems to require a prolonged activation of Akt1 and is distinct from the role of this kinase as a signaling molecule that is quickly and transiently activated in response to growth factors. This new function of Akt1 is independent of eNOS activation, as we observed no changes in TSPs levels in Nos3−/− mice. TSPs are known to inhibit neovascularization by several mechanisms, including induction of endothelial cell apoptosis through interaction with CD36 (ref. 28), inhibition of metalloproteinase activity, which in turn might prevent mobilization of endothelial cells29, and inhibition of cell-cycle progression30. The reduced levels of TSPs at the sites of angiogenesis in Akt1−/− mice could account for a lack of endothelial cell apoptosis, a normal proliferation rate of endothelial cells in vivo, and increased angiogenesis in our in vivo models. Our finding that excessive angiogenesis was corrected by re-expression of TSP-1 and TSP-2 in the skin of Akt1−/− mice strongly supports a causal role for TSPs in this response.
In addition to their well-known effects on endothelial cell proliferation and apoptosis, our study emphasizes another notable function of TSPs, namely the maintenance of vascular integrity. The leaky vascular phenotype observed in TSP-2-deficient mice31 also supports our conclusions. Finally, the reduced collagen content in Akt1−/− mice could be another consequence of the lack of TSPs, which are known to regulate collagen matrix assembly32.
In our experimental models, impaired vascular maturation but not other aspects of vascular development seem to be related to the eNOS pathway in Akt1−/− mice. This finding is consistent with substantial evidence indicating that eNOS is not an exclusive target of Akt and, therefore, a complete overlap of phenotypes in Akt1−/− and Nos3−/− mice is not anticipated33. Indeed, recent studies using mice that overexpress Akt emphasize the role of other Akt effectors, such as mTOR, in the regulation of angiogenesis in cardiac tissues9. Thus, although eNOS undoubtedly represents an important component of the Akt signaling network in endothelial cells, other mechanisms contribute to both the pro- and antiangiogenic functions of Akt. It is also important to consider that different animal models and tissue environments might show different consequences of Akt signaling.
In summary, our study shows that Akt1 is vital for the regulation of vascular permeability, angiogenic responses and subsequent vascular maturation. Our results and recent reports9,34,35 suggest that Akt activity requires fine-tuning in a time-dependent manner. It is also clear that because of its numerous downstream targets, Akt1 affects several regulatory pathways of the vascularization response, including pro- as well as antiangiogenic components. In our models using Akt1−/− mice, changes in extracellular matrix and reduction in the levels of TSP-1 and TSP-2, endogenous inhibitors of angiogenesis, shifted the balance toward increased angiogenesis. As neoangiogenesis and vascular leakage are important pathophysiological features of cancer, ischemic injury, inflammation and degenerative disorders, our findings have broad therapeutic implications.
METHODS
Animals
We generated Akt1−/− mice as previously described6 and maintained them in the 129 R1/C57BL/6 background. We used sex-matched 8–12-week-old wild-type and Akt1−/− littermate mice for the study. We performed all procedures according to protocols approved by Cleveland Clinic Foundation Institutional Animal Care and Use Committee.
Aortic ring assay and isolation of endothelial cells
We performed aortic ring assays as previously described36. We stimulated vessel sprouting using 30 ng/ml of VEGF for 14 d. We harvested endothelial cells outgrown from rings and quantified them by FACS analysis using CD31-specific antibody. We used the same protocol for the isolation of endothelial cells. We isolated microvascular endothelial cells from lungs as previously described37. For isolation of endothelial cells from Matrigel implants, we used a similar protocol.
Western blot analysis
We lysed tissue and tumor samples and performed SDS-PAGE and western blotting as described previously18.
Analysis of endothelial cells in vitro
We performed cell migration assays as previously described38,39 using Transwell plates. We incubated endothelial cells from wild-type or Akt1−/− mice with or without VEGF for 4 h at 37 °C. We microscopically quantified migration (magnification, ×200). We performed analyses of fibrinogen binding, integrin and TSP expression by FACS as previously described40.
Matrigel plug assay
We performed Matrigel plug assay as previously described36. On day 5, we removed Matrigel plugs and digested them using 5 ml Drabkin reagent (Sigma) and quantified neovessel formation using a hemoglobin assay according to the manufacturer's protocol. In a parallel experiment, we used the harvested Matrigel implants to isolate endothelial cells.
Tumor angiogenesis
We performed tumor angiogenesis as previously described36 using B16F10 melanoma cells, and collected the tumors 7 d after injection as indicated. We processed tissue sections as previously described41.
Expression and analyses of Ad-VEGF-A165,Ad-VEGF-DΔNΔC and Ad-VEGF-D in vivo
We stimulated angiogenesis in skin using adenoviruses carrying genes encoding VEGF-A165, VEGF-DΔNΔC, VEGF-D or green fluorescence protein (GFP) as negative control as previously described13. Alternatively, we injected recombinant VEGF-A165, VEGF-D and VEGF-DΔNΔC subcutaneously along with 0.1 ml of Matrigel. We isolated skin along with the muscle layer 7 d after injection and processed samples for immunohistochemistry. The means of von Willebrand factor (vWF)-positive vascular area were 1.5% and 1.9% for wild-type and Akt1−/− control samples, respectively.
Bone marrow transplantation
We lethally irradiated both wild-type and Akt1−/− 6–8-week-old recipient mice (n = 11 per group), and we transplanted 1 × 107 bone marrow cells from wild-type littermates by tail vein injection as previously described42.
Immunohistochemistry and image analysis
We analyzed vessel density and area as described previously13, based on vWF, CD31 and laminin staining. We detected the expression of TSP-1, vWF, CD105, fibrin and SMA using antibodies from Abcam, Dako, BD Bioscienses, Accurate Chemical and Scientific Corporation and Sigma, respectively, in tissue sections by immunohistochemistry (Supplementary Methods online). We quantified the area of TSP-1 expression and plasma leakage using Image Pro-Plus software as previously described39.
Vascular permeability assay in vivo
We performed this assay using Evans blue dye (30 mg/kg in 100 μl PBS; Sigma) as previously described43 with modifications (Supplementary Methods online). We performed a characterization of microvessel morphology and leakage sites in ricin lectin-stained whole mounts of ear skin using a previously described protocol15 with modifications (Supplementary Methods online).
Vascular permeability assay in vitro
We performed a vascular cell mono-layer permeability assay of primary endothelial cells isolated from wild-type and Akt1−/− animals as described in44 with modifications (Supplementary Methods online).
Thbs2 promoter activity assay
We transiently transfected human umbilical vein endothelial cells (HUVECs) with a vector encoding luciferase under the control of the Thbs2 promoter (starting at nucleotide −2,701) as described45 or cotransfected them with retroviral constructs encoding dominant negative Akt1 (Akt1K179M) or myrAkt1 with vehicle as control. We determined luciferase activity in cell lysates with a luciferase assay kit (Promega) using a Wallac Victor2 1420 Multilabel counter (Perkin Elmer) and normalized the result to total protein as measured by a protein assay kit (Bio-Rad).
Statistical analysis
All data are presented as means ± s.e.m. for in vivo assays or means ± s.d. for ex vivo studies and were analyzed using SPSS software (Norusis SPSS Inc., Release 6.0). We performed all analyses using two-sample t-tests with a two-tailed significance set at the 0.05 level. Before any test, we examined normality of the data and performed appropriate transformation when the normal distribution assumption was not met. We reported the actual P value for each test. In addition, we examined the relationship between TSP-1 expression intensity and vessel density using Pearson correlation coefficient with a significance level of 0.01.
ACKNOWLEDGMENTS
We acknowledge support from the US National Institutes of Health (HL071625 and DK060933 to T.V.B. and CA90764 to N.H). We thank J. Li for his help in statistical analysis, J. Drazba, A. Vasantji and Imaging Core, Cleveland Clinic Foundation for the help in tissue processing and image analysis and L. Mavrakis and S. Bundy for supply of HUVEC, harvested through Birthing Services Department at the Cleveland Clinic Foundation and Perinatal Clinical Research Center at the MetroHealth Hospital.
Footnotes
Note: Supplementary information is available on the Nature Medicine website.
COMPETING INTERESTS STATEMENT
The authors declare that they have no competing financial interests.
References
Full text links
Read article at publisher's site: https://doi.org/10.1038/nm1307
Read article for free, from open access legal sources, via Unpaywall:
https://europepmc.org/articles/pmc2277080?pdf=render
Citations & impact
Impact metrics
Citations of article over time
Alternative metrics
Smart citations by scite.ai
Explore citation contexts and check if this article has been
supported or disputed.
https://scite.ai/reports/10.1038/nm1307
Article citations
Exploring the mechanism of Zhengxintai Formula for the treatment of coronary heart disease based on network pharmacology.
Medicine (Baltimore), 103(41):e40065, 01 Oct 2024
Cited by: 0 articles | PMID: 39465849 | PMCID: PMC11479439
Identification of potential miRNA-mRNA regulatory network contributing to pathogenesis of polycystic ovarian syndrome.
Ir J Med Sci, 06 Sep 2024
Cited by: 0 articles | PMID: 39240275
Muscone abrogates breast cancer progression through tumor angiogenic suppression via VEGF/PI3K/Akt/MAPK signaling pathways.
Cancer Cell Int, 24(1):214, 20 Jun 2024
Cited by: 0 articles | PMID: 38898449 | PMCID: PMC11188526
Increase in Vascular Endothelial Growth Factor (VEGF) Expression and the Pathogenesis of iMCD-TAFRO.
Biomedicines, 12(6):1328, 14 Jun 2024
Cited by: 1 article | PMID: 38927535
Review
Value analysis of ITLN1 in the diagnostic and prognostic assessment of colorectal cancer.
Transl Cancer Res, 13(6):2877-2891, 25 Jun 2024
Cited by: 1 article | PMID: 38988920
Go to all (295) article citations
Other citations
Data
Similar Articles
To arrive at the top five similar articles we use a word-weighted algorithm to compare words from the Title and Abstract of each citation.
Thrombospondin-4 in tissue remodeling.
Matrix Biol, 75-76:300-313, 11 Nov 2017
Cited by: 45 articles | PMID: 29138119 | PMCID: PMC6005712
Review Free full text in Europe PMC
Akt1/protein kinase Balpha is critical for ischemic and VEGF-mediated angiogenesis.
J Clin Invest, 115(8):2119-2127, 01 Aug 2005
Cited by: 295 articles | PMID: 16075056 | PMCID: PMC1180542
Thrombospondins as anti-angiogenic therapeutic agents.
Curr Pharm Des, 9(7):583-588, 01 Jan 2003
Cited by: 31 articles | PMID: 12570805
Review
Thrombospondin 2 expression is correlated with inhibition of angiogenesis and metastasis of colon cancer.
Br J Cancer, 79(2):354-359, 01 Jan 1999
Cited by: 74 articles | PMID: 9888480 | PMCID: PMC2362194
Funding
Funders who supported this work.
NCI NIH HHS (2)
Grant ID: R01 CA090764
Grant ID: CA90764
NHLBI NIH HHS (2)
Grant ID: HL071625
Grant ID: R01 HL071625
NIDDK NIH HHS (2)
Grant ID: DK060933
Grant ID: R01 DK060933
