Fulvene-5 potently inhibits NADPH oxidase 4 and blocks the growth of endothelial tumors in mice

Hemangiomas are the most common type of tumor in infants. As they are endothelial cell–derived neoplasias, their growth can be regulated by the autocrine-acting Tie2 ligand angiopoietin 2 (Ang2). Using an experimental model of human hemangiomas, in which polyoma middle T–transformed brain endothelial (bEnd) cells are grafted subcutaneously into nude mice, we compared hemangioma growth originating from bEnd cells derived from wild-type, Ang2^+/–, and Ang2^–/– mice. Surprisingly, Ang2-deficient bEnd cells formed endothelial tumors that grew rapidly and were devoid of the typical cavernous architecture of slow-growing Ang2-expressing hemangiomas, while Ang2^+/– cells were greatly impaired in their in vivo growth. Gene array analysis identified a strong downregulation of NADPH oxidase 4 (Nox4) in Ang2^+/– cells. Correspondingly, lentiviral silencing of Nox4 in an Ang2-sufficient bEnd cell line decreased Ang2 mRNA levels and greatly impaired hemangioma growth in vivo. Using a structure-based approach, we identified fulvenes as what we believe to be a novel class of Nox inhibitors. We therefore produced and began the initial characterization of fulvenes as potential Nox inhibitors, finding that fulvene-5 efficiently inhibited Nox activity in vitro and potently inhibited hemangioma growth in vivo. In conclusion, the present study establishes Nox4 as a critical regulator of hemangioma growth and identifies fulvenes as a potential class of candidate inhibitor to therapeutically interfere with Nox function.

Hemangiomas are the most common neoplasms of infancy. While the exact pathogenesis of these lesions remains obscure, prior studies have shown that they are clonal neoplasms (1). These lesions, unlike most other neoplasms, undergo spontaneous regression. The molecular repertoire of angiogenic cytokines controlling hemangioma growth is poorly understood, but evidence has emerged that the Tie2 ligand angiopoietin 2 (Ang2) plays a dominant role in hemangioma growth (1, 2). While hemangiomas of infancy are exclusively a disorder of childhood, a histologically similar lesion is acquired through infection by the bacterium Bartonella bacilliformis in adults. B. bacilliformis infection similarly leads to increased production of Ang2 (3).

We have previously shown that pharmacologic blockade of Ang2 in a murine model of hemangiomas leads to decreased growth in vivo (2). We therefore expected that polyoma middle T oncogene–transformed endothelial cells derived from Ang2-deficient mice would be unable to form hemangiomas in vivo. Surprisingly, Ang2-deficient endothelioma cells formed tumors much faster than Ang2-expressing endothelioma cells, but these tumors were not hemangiomas; rather, they histologically resembled angiosarcomas. We therefore hypothesized that endothelioma-derived Ang2 may control critical downstream mediators of hemangioma growth. Exploratory gene array experiments identified the NADPH oxidase Nox4 as being downregulated in Ang2 heterozygous cells that grow poorly in vivo. Thus, loss of Ang2 appears to affect the ability of endothelial cells to form hemangiomas, while loss of Nox4 appears to have a more severe phenotype, in significantly inhibiting the growth of hemangiomas in vivo. We also demonstrate that novel fulvenes, which block Nox4, have potent inhibitory effects on hemangioma growth in vivo.

Ang2^–/– cells form rapidly growing tumors in vivo while Ang2^+/– and Ang2^+/+ form tumors with greatly delayed latency. Polyoma middle T–transformed endothelial cells were established from primary endothelial cells isolated from P3 murine brains of wild-type, Ang2-heterozygous, and Ang2-deficient mice. In order to study the tumor-forming potential of these cells, low numbers of Ang2^–/–, Ang2^+/–, and Ang2^+/+ cells (2.5 × 10⁵) were injected subcutaneously into nude mice. The mice were observed over a period of 4 weeks. Under these experimental conditions, Ang2^+/+ and Ang2^+/– cells formed slow-growing tumors with the characteristic morphological appearance of cavernous hemangiomas (Figure 1, A–C). In contrast, Ang2^–/– cells generated from knockout littermates formed much faster–growing tumors that had lost the typical cavernous hemangioma-like morphology of tumors originating from wild-type cells and resembled angiosarcoma instead (Figure 1D). Immunostaining for CD31, RT-PCR for CD31, and CD31 immunohistochemistry demonstrated that all lesions retained endothelial differentiation, but tumors derived from Ang2^–/– cells were histologically sarcomatous rather than hemangiomas. In fact, the growth difference was so pronounced that a detailed side-by-side comparison of wild-type hemangiomas with heterozygous and Ang2^–/– hemangiomas was not possible because Ang2^–/– hemangioma-bearing mice had to be sacrificed prior to the growth of Ang2^+/– or Ang2^+/+ neoplasms.

Figure 1

Differential tumor growth in endotheliomas derived from Ang2 wild-type, heterozygous, and knockout mice. Tumor formation in Ang2^–/– cells and Ang2^+/– cells in vivo (A). Six mice were injected with 2.5 × 10⁵Ang^–/– or Ang^+/– cells. Animals were euthanized on day 30, secondary to tumor burden in the Ang^–/– animals. The figure depicts the average tumor volume (mm³) in both experimental groups. Error bar represents SEM. P < 0.05 versus control. Morphological analysis revealed typical cavernous hemangiomas in wild-type endothelioma–derived hemangiomas (B) and in Ang2^+/– endothelioma–derived hemangiomas (C) compared with the solid growth of Ang2^–/– endothelioma–derived sarcomas (D). (E–G) CD31 immunohistochemistry of Ang2^+/+, Ang2^+/–, and Ang2^–/– tumors, respectively, with positive staining in all 3 tumors, but decreased differentiation in the Ang2^–/– tumors. (H–J) Immunocytochemistry of CD31, with pink staining representing CD31 and blue DAPI staining representing nuclei. (J) RT-PCR of Ang2^+/+, Ang2^+/–, and Ang2^–/– cells for CD31, with GAPDH as a loading control. Original magnification, ×40. (K) Represented is RT-PCR for CD31 in all 3 phenotypes of endothelial cells. Note presence of the endothelial marker CD31 in all 3 types of cells. GAPDH represents a loading control. PCR was performed using QIAGEN reagents per manufacturer’s protocol and 35 cycles performed at an annealing temperature of 55°C. To confirm expression of the endothelial marker CD31, PCR was performed using the following primer sequences: forward, 5′-GTGAAGGTGCATGGCGTATC-3′; reverse 5′-CACAAAGTTCTCGTTGGAGGT-3′, producing a 192-bp product. PCR for GAPDH was used as a control: forward, 5′-AGGTCGGTGTGAACGGATTT-3′; reverse, 5′-CTCCTGGAAGATGGTGATGG-3′, resulting in a 224-bp band.

Nox4 and Nrarp are strongly downregulated in Ang2-heterozygous endothelial cells. Comparative gene array experiments were performed in order to yield mechanistic insight into the pronounced growth difference between Ang2-deficient and Ang2-heterozygous endothelial cells. Among the differentially expressed genes, Notch-related ankyrin repeat protein (Nrarp) and Nox4 were identified as among the most differentially expressed. Confirmatory quantitative RT-PCR analyses validated the array experiments and demonstrated the pronounced downregulation of Nrarp and Nox4 expression in Ang2^+/– cells compared with Ang2^–/– cells (Figure 2).

Figure 2

Downregulation of Nrarp and Nox4 mRNA in Ang2 heterozygous endothelioma cells. RT-PCR analysis of Ang2^–/– and Ang2^+/– cells. Nrarp mRNA expression (A) and Nox4 mRNA expression (B) are increased in Ang2^–/– cells compared with Ang2^+/– cells. Bars represent the average of triplicate experiments, and error bars indicate SEM. P < 0.05 versus control.

bEnd.3 Nox4shRNA cells show decreased expression of Ang2 and Nox4 in vitro. In order to determine whether Ang2 and Nox4 expression is associatively linked, Nox4 shRNA–mediated knockdown cells were generated in brain endothelial 3 (bEnd.3) cells, a well-established model of hemangiomas. Knockdown was efficient, with a 91% decrease in Nox4 expression (Figure 3A) and a concomitant 73% decrease in Ang2 expression in bEnd.3 Nox4shRNA cells (Figure 3B) compared with control cells.

Figure 3

Nox4 siRNA downregulates endogenous Nox4 and Ang2 in bEnd.3 cells. Expression of Nox4 (A) and Ang2 (B) in bEnd.3 Nox4siRNA cells. Nox4 mRNA and Ang2 mRNA expression are decreased in bEnd.3 Nox4shRNA cells. Bars represent the average of triplicate experiments, and error bars indicate SEM. P < 0.05 versus control.

bEnd.3 Nox4shRNA cells show reduced tumor growth. In order to determine whether the high tumorigenic potential of Ang2^–/– endothelium may be related to dysregulated Nox4 expression, Nox4 expression was shRNA silenced in bEnd.3 endothelioma cells and the tumorigenicity of Nox4-silenced and control bEnd.3 cells was studied in vivo. Toward this end, large numbers of bEnd.3 cells (10⁶) were injected subcutaneously into nude male mice and tumor growth was monitored for a period of 2 weeks. Control bEnd.3 cells readily formed large hemangiomas, whereas Nox4-silenced cells grew only small nodules (Figure 4).

Figure 4

Tumor volumes in bEnd.3 control versus bEnd.3 Nox4shRNA. Six mice were injected with 10⁶ bEnd.3 control and bEnd.3 Nox4shRNA cells. Animals were euthanized on day 14, secondary to tumor burden in the bEnd.3 control animals. The figure depicts average tumor volume (mm³) in each group. Error bars represent SEM. P < 0.05 versus control.

Fulvene-5 inhibits Nox function in vitro. In order to inhibit Nox4 function pharmacologically, we tested fulvene derivatives as a potentially new class of Nox inhibitors. We used HEK293 cells stably transfected with constitutively active Nox4 and COSphox cells harboring inducible Nox2/p47phox/p67phox complex and measured hydrogen peroxide generation by homovanillic acid (HVA) assay. Fulvene-5 treatment decreased Nox4- and Nox2-mediated ROS production by approximately 40% (Figure 5).

Figure 5

Fulvene-5 inhibits ROS generation by Nox4 and Nox2 in vitro. 293 cells were treated with vehicle control or 5 μM indigo fulvene as described in Methods. COSphox cells were additionally either left unstimulated or stimulated with phorbol ester (PMA). H₂O₂ production was measured after 60 minutes. Results are shown as a percentage relative to untreated control (100%) and are the mean of n = 4 (Nox2) or depict a representative experiment (Nox4, n = 2).

Fulvene-5 downregulates Ang2, Nrarp, Dll4, and PlGF expression in bEnd.3 cells. We confirmed the decreased expression of Ang2, delta-like ligand 4 (DLL4), placental growth factor (PLGF), and Nrarp in a dose-dependent manner in bEnd.3 cells treated with fulvene-5 and determined a 99% decrease in Nrarp expression, a 60% decrease in PLGF expression, a 75% decrease in DLL4 expression, and at the same time, a 25% decrease in Ang2 expression in bEnd.3 treated with 1.0 μM of fulvene-5, consistent with data obtained from either loss of Ang2 or Nox4 knockdown in polyoma-transformed endothelial cells (Figure 6).

Figure 6

Relative expression levels of Ang2, Dll4, and PlGF in bEnd.3 cells treated with fulvene-5. Relative expression levels indicate a dose-dependent decrease. Bars represent the average of triplicate experiments (corrected for 18S RNA), and error bars indicate the SEM. *P < 0.05 versus control.

Fulvene-5 reduces hemangioma formation in mice injected with bEnd.3 cells. To determine whether fulvene-5–induced loss of Nox4 demonstrated antitumorigenicity, we injected nude mice subcutaneously with high numbers of bEnd.3 cells (10⁶). Mice were treated with fulvene-5 (120 mg/kg/d) or vehicle control for 2 weeks, and tumor growth as well as tumor burden were analyzed thereafter. Tumor growth in mice treated with fulvene-5 was significantly reduced compared with that in vehicle control–treated mice (Figure 7, A and B).

Figure 7

Treatment with vehicle control and fulvene-5 in vivo. For each treatment condition, 3 mice were subcutaneously injected on day 1 with 10⁶ bEnd.3 cells and monitored for tumor development. From day 2 to day 12, mice were treated with fulvene-5 by making a stock solution of 4 mg fulvene-5 compound dissolved in 100 μl of 100% ethanol plus 1.1 ml of intralipid and then sonicated for 15 seconds. The treatment stock solution (330 μl) was injected intraperitoneally into each of 3 mice of the fulvene-5 treatment group. For control animals, a stock solution was made of 100 μl of 100% ethanol plus 1.1 ml of intralipid solution, and 330 μl of stock solution was injected intraperitoneally into each of 3 mice of the control group. Injections were continued for 2 weeks. On day 9 and day 12, tumors were measured in all 6 animals, and there was a significant difference in tumor volume between the control and fulvene-5–treated mice. No toxicity was noted following injection. Animals were euthanized on day 14, secondary to tumor burden in the control animals. Photos represent average tumor burden in each of the 2 groups (A), and tumor volume (mm³) is graphically depicted (B). Error bars represent SEM. P < 0.05 versus control.

The Tie2 ligand Ang2 has been identified as an antagonistic ligand of constitutive Ang1/Tie2 signaling (4). As such, it destabilizes the vascular endothelium and primes it to respond to exogenous cytokines (5). Yet Ang2 functions are contextual, and Ang2 has been shown to be capable of acting both as an inhibitory ligand of Tie2 and as a stimulating ligand even though the molecular mechanisms controlling agonistic versus antagonistic functions of Ang2 are largely unknown (5). Importantly, Ang2 is almost exclusively produced by endothelial cells themselves (6, 7). Since it is produced by its own target cell, it functions as an autocrine-acting growth factor of endothelial cells.

Hemangiomas are endothelial-derived benign vascular neoplasms. Corresponding to its endothelial cell origin, Ang2 has in recent years been identified as a major regulator of hemangioma growth (1, 2). Based on these findings, the present study was aimed at molecularly dissecting the role of Ang2 during hemangioma growth. The study yielded surprising findings that shed mechanistic insights into the pathogenesis of hemangioma growth and open the door toward modalities that may therapeutically interfere with hemangiomas. Specifically, the present study identifies (a) loss of hemangioma differentiation in tumors derived from Ang2-knockout cells, indicating a requirement for Ang2 in hemangioma genesis; (b) the downregulation of Nox4 in polyoma-transformed endothelial cells in poorly tumorigenic Ang2^+/– cells; (c) reduced hemangioma Ang2 production upon Nox4 silencing; (d) reduced hemangioma growth upon Nox4 silencing; and (e) fulvene-5 as inhibitor of Nox4 and correspondingly as potent inhibitor of hemangioma growth.

NADPH oxidases (Nox) are a family of NADPH-dependent enzymes that generate reactive oxygen species and consist of 7 members in humans (8–13). These enzymes are part of multiunit complexes that may include p22phox, cytosolic oxidase components, and the small GTPases Rac1 and Rac2. The first Nox enzyme to be identified was Nox2 (gp91phox), which is frequently mutated in chronic granulomatous disease and is responsible for the phagocyte respiratory burst (14, 15). Other Nox/Duox enzymes were discovered as superoxide/hydrogen peroxide–generating enzymes in other cell types. Nox1 is highly expressed in gastrointestinal epithelium, Nox3 in the inner ear, and Nox4 in a number of tissues including malignant melanoma (16, 17). We have previously shown that Nox4 is expressed in endothelial cells and in bEnd.3 cells (2).

A role for Nox in hemangioma growth was circumstantially suggested in earlier experiments. We have demonstrated that topical application of eosin, a potent Nox inhibitor, causes regression of ulcerated hemangiomas of infancy (18). Likewise, systemic propranolol, a beta blocker, has also shown efficacy in hemangiomas of infancy (18). Propranolol has activity against reactive oxygen in addition to its activity against the β-adrenergic receptor (19).

The identification of a critical role of Nox4 in hemangioma growth suggested that Nox may serve as a target to therapeutically interfere with hemangioma growth. Small molecule inhibitors of Nox have been synthesized, but in vivo data are limited. Based upon structural similarities to the well-studied Nox inhibitor diphenylene iodonium, we found that triphenylmethane dyes are potent inhibitors of Nox. Unfortunately, many of these compounds are poorly tolerated in vivo. The structure of triphenylmethanes indicated a need for conjugated aromatic structures that allow electron delocalization. In order to create similar structures without the triphenylmethane backbone, we synthesized fulvenes through the addition of sodium cyclopentadienide to ketones. Fulvenes are aromatic structures, yet they are highly water soluble. By conjugation of sodium cyclopentadienide with the ketone indigo, we synthesized fulvene-5, which showed inhibitory activity against both Nox4 and Nox2 function. We found that fulvene-5 blocked the production of Ang2 as well as other genes downstream of Nox4, including Nrarp. In addition, fulvene-5 potently inhibited the growth of bEnd.3 hemangioma in vivo. This represents what we believe is the first pharmacologic use of any compound with the fulvene structure.

The detailed molecular mechanisms of Nox4 function during hemangioma growth remain to be elucidated. bEnd.3 hemangiomas are a classic model of endothelial cell recruitment in that they grow through a process called host recruitment (20, 21). Increased superoxide generation and increased production of Ang2 may contribute to host recruitment including the recruitment of endothelial precursor cells. Traps devised against the VEGF-induced Notch ligand Dll4 have been produced to address the resistance of tumors to anti-VEGF therapies (22). Blockade of Nox enzymes with small molecules may be an attractive strategy beyond their application in hemangiomas because it may block both Dll4 signaling and superoxide-dependent HIF2α-induced VEGF (23). Combination or sequential therapies using VEGF blockade and Nox4 blockade using novel small molecules such as fulvenes may result in improved responses for angiogenic disorders including cancer.

J.L. Arbiser was supported by NIH grant R01 AR02030, and grants from the Jamie Rabinowitch-Davis Foundation and the Minsk Foundation. B. Govindarajan was supported by a Dermatology Foundation Career Development award. H.G. Augustin and Y. Reiss were supported by the Deutsche Forschungsgemeinschaft (SFB-TR23, Vascular Differentiation and Remodeling).

Address correspondence to: Jack L. Arbiser, Department of Dermatology, Emory University School of Medicine, WMB 5309, 1639 Pierce Drive, Atlanta, Georgia 30322, USA. Phone: (404) 727-5063; Fax: (404) 727-0923; E-mail: jarbise@emory.edu.

Conflict of interest: A patent is pending on fulvene-5, with J.L. Arbiser as inventor and Emory University as owner of the patent.

Nonstandard abbreviations used: Ang2, angiopoietin 2; bEnd, brain endothelial; DLL4, delta-like ligand 4; Nox4, NADPH oxidase 4; Nrarp, Notch-related ankyrin repeat protein; PLGF, placental growth factor.

Reference information: J. Clin. Invest.119:2359–2365 (2009). doi:10.1172/JCI33877

1. Yu, Y., Varughese, J., Brown, L.F., Mulliken, J.B., Bischoff, J. 2001. Increased Tie2 expression, enhanced response to angiopoietin-1, and dysregulated angiopoietin-2 expression in hemangioma-derived endothelial cells. Am. J. Pathol. 159:2271-2280.
   View this article via: PubMed Google Scholar
2. Perry, B.N., et al. 2006. Pharmacologic blockade of angiopoietin-2 is efficacious against model hemangiomas in mice. J. Invest Dermatol. 126:2316-2322.
   View this article via: CrossRef PubMed Google Scholar
3. Cerimele, F., et al. 2003. Infectious angiogenesis: Bartonella bacilliformis infection results in endothelial production of angiopoetin-2 and epidermal production of vascular endothelial growth factor. Am. J. Pathol. 163:1321-1327.
   View this article via: PubMed Google Scholar
4. Maisonpierre, P.C., et al. 1997. Angiopoietin-2, a natural antagonist for Tie2 that disrupts in vivo angiogenesis. Science. 277:55-60.
   View this article via: CrossRef PubMed Google Scholar
5. Augustin, H.G., Koh, G.Y., Thurston, G., Alitalo, K. 2009. Control of vascular morphogenesis and homeostasis through the angiopoietin-Tie system. Nat. Rev. Mol. Cell Biol. 10:165-177.
   View this article via: CrossRef PubMed Google Scholar
6. Fiedler, U., et al. 2004. The Tie-2 ligand angiopoietin-2 is stored in and rapidly released upon stimulation from endothelial cell Weibel-Palade bodies. Blood. 103:4150-4156.
   View this article via: CrossRef PubMed Google Scholar
7. Stratmann, A., Risau, W., Plate, K.H. 1998. Cell type-specific expression of angiopoietin-1 and angiopoietin-2 suggests a role in glioblastoma angiogenesis. Am. J. Pathol. 153:1459-1466.
   View this article via: PubMed Google Scholar
8. Arbiser, J.L., et al. 2002. Reactive oxygen generated by Nox1 triggers the angiogenic switch. Proc. Natl. Acad. Sci. U. S. A. 99:715-720.
   View this article via: CrossRef PubMed Google Scholar
9. Bedard, K., Krause, K.H. 2007. The NOX family of ROS-generating NADPH oxidases: physiology and pathophysiology. Physiol. Rev. 87:245-313.
   View this article via: CrossRef PubMed Google Scholar
10. Geiszt, M. 2006. NADPH oxidases: new kids on the block. Cardiovasc. Res. 71:289-299.
    View this article via: CrossRef PubMed Google Scholar
11. Kawahara, B.T., Quinn, M.T., Lambeth, J.D. 2007. Molecular evolution of the reactive oxygen-generating NADPH oxidase (Nox/Duox) family of enzymes. BMC Evol. Biol. 7:109.
    View this article via: PubMed Google Scholar
12. Kwon, J., et al. 2004. Reversible oxidation and inactivation of the tumor suppressor PTEN in cells stimulated with peptide growth factors. Proc. Natl. Acad. Sci. U. S. A. 101:16419-16424.
    View this article via: CrossRef PubMed Google Scholar
13. Yu, L., Zhen, L., Dinauer, M.C. 1997. Biosynthesis of the phagocyte NADPH oxidase cytochrome b558. Role of heme incorporation and heterodimer formation in maturation and stability of gp91phox and p22phox subunits. J. Biol. Chem. 272:27288-27294.
    View this article via: CrossRef PubMed Google Scholar
14. Dinauer, M.C., Orkin, S.H., Brown, R., Jesaitis, A.J., Parkos, C.A. 1987. The glycoprotein encoded by the X-linked chronic granulomatous disease locus is a component of the neutrophil cytochrome b complex. Nature. 327:717-720.
    View this article via: CrossRef PubMed Google Scholar
15. Parkos, C.A., Dinauer, M.C., Jesaitis, A.J., Orkin, S.H., Curnutte, J.T. 1989. Absence of both the 91kD and 22kD subunits of human neutrophil cytochrome b in two genetic forms of chronic granulomatous disease. Blood. 73:1416-1420.
    View this article via: PubMed Google Scholar
16. Brar, S.S., et al. 2001. Reactive oxygen species from NAD(P)H:quinone oxidoreductase constitutively activate NF-kappaB in malignant melanoma cells. Am. J. Physiol. Cell Physiol. 280:C659-C676.
    View this article via: PubMed Google Scholar
17. Govindarajan, B., et al. 2007. Overexpression of Akt converts radial growth melanoma to vertical growth melanoma. J. Clin. Invest. 117:719-729.
    View this article via: JCI CrossRef PubMed Google Scholar
18. Lapidoth, M., et al. 2009. Efficacy of topical application of eosin for ulcerated hemangiomas. J. Am. Acad. Dermatol. 60:350-351.
    View this article via: CrossRef PubMed Google Scholar
19. Lodha, S., et al. 2002. Angiotensin II-induced mesangial cell apoptosis: role of oxidative stress. Mol. Med. 8:830-840.
    View this article via: PubMed Google Scholar
20. Bautch, V.L., Toda, S., Hassell, J.A., Hanahan, D. 1987. Endothelial cell tumors develop in transgenic mice carrying polyoma virus middle T oncogene. Cell. 51:529-537.
    View this article via: CrossRef PubMed Google Scholar
21. Williams, R.L., et al. 1989. Endothelioma cells expressing the polyoma middle T oncogene induce hemangiomas by host cell recruitment. Cell. 57:1053-1063.
    View this article via: CrossRef PubMed Google Scholar
22. Noguera-Troise, I., et al. 2006. Blockade of Dll4 inhibits tumour growth by promoting non-productive angiogenesis. Nature. 444:1032-1037.
    View this article via: CrossRef PubMed Google Scholar
23. Maranchie, J.K., Zhan, Y. 2005. Nox4 is critical for hypoxia-inducible factor 2-alpha transcriptional activity in von Hippel-Lindau-deficient renal cell carcinoma. Cancer Res. 65:9190-9193.
    View this article via: CrossRef PubMed Google Scholar
24. Gale, N.W., et al. 2002. Angiopoietin-2 is required for postnatal angiogenesis and lymphatic patterning, and only the latter role is rescued by Angiopoietin-1. Dev. Cell. 3:411-423.
    View this article via: CrossRef PubMed Google Scholar
25. Kiefer, F., et al. 1994. Endothelial cell transformation by polyomavirus middle T antigen in mice lacking Src-related kinases. Curr. Biol. 4:100-109.
    View this article via: CrossRef PubMed Google Scholar
26. Li, J., et al. 2006. The NADPH oxidase NOX4 drives cardiac differentiation: role in regulating cardiac transcription factors and MAP kinase activation. Mol. Biol. Cell. 17:3978-3988.
    View this article via: CrossRef PubMed Google Scholar
27. Dull, T., et al. 1998. A third-generation lentivirus vector with a conditional packaging system. J. Virol. 72:8463-8471.
    View this article via: PubMed Google Scholar
28. Chajara, K., Ottosson, H. 2004. An improved pathway to 6,6-disubstituted fulvenes. Tetrahedron Lett. 45:6741-6744.