Tumor angiogenesis is recognized as an essential process in the development of malignant tumors. Advances in this area of research have focused attention on angiogenesis inhibitors as potential novel therapeutic agents for cancer, while studies on the biology of tumor-derived angiogenic growth factors are increasing our understanding of blood vessel formation from early embryogenesis onward.

The annual W. Alton Jones Cell Science Center Symposium on Cellular Endocrinology was held September 11-14, 1997, in Lake Placid, New York, in the heart of the Adirondack Park. This year's topic was "Cell Signaling and Tumor Angiogenesis."

Judah Folkman (Children's Hospital and Harvard University Medical School, Boston) delivered the keynote address. Over the past 25 years Folkman and his colleagues have championed the hypothesis that the survival and development of malignant solid tumors requires that growing tumor masses become vascularized. This tumor vascularization occurs through angiogenesis, the formation of new blood vessels from surrounding, preexisting vessels, which grow into developing tumors in a directed manner. By making this connection to the vascular system, tumors obtain nutrients and oxygen, which allow continued growth; they receive signals from endothelial cell-derived paracrine factors; and they gain a potential escape route for metastasizing to other regions of the body.

Over the years this hypothesis has received support from several experimental findings. Isolated tumor cells in culture invariably produce angiogenic growth factors such as fibroblast growth factor 1 (acidic FGF), FGF-2 (basic FGF), or vascular endothelial growth factor/vascular permeability factor (VEGF/VPF). The transition of immortal but non-tumorigenic cells to a malignant tumorigenic phenotype is accompanied by the ability to release angiogenic factors. And specific inhibitors of angiogenic growth factor activity, such as neutralizing antibodies to VEGF or dominant-negative VEGF receptor mutants, inhibit tumor growth in experimental mice. It was against this background that the symposium was held.

A number of angiogenic growth factors are known, but outside of wound healing and the female reproductive cycle, angiogenesis normally occurs infrequently in adult vertebrates. As a consequence, the endothelial cells that line large blood vessels and comprise microvessels are generally quiescent.

The situation is quite different during embryogenesis, in which it is vital for the developing embryo to generate an extensive vasculature and blood cells de novo within a short period of time. Furthermore, in mammalian development the fetal circulatory system must make a connection to the maternal circulatory system through the placenta. Robert Auerbach (Laboratory of Developmental Biology, University of Wisconsin) described the differentiation of endothelial cells and blood cells from common precursor cells, hemangioblasts, in the yolk sac of the developing mouse embryo. Yolk sac endothelial cells are able to induce the differentiation of hematopoietic stem cells, and they can contribute to the vasculature of experimental tumors. The latter observation raises the possibility that tumor blood vessels may arise from the differentiation of migrating angioblasts in addition to the proliferation and outgrowth of endothelial cells from existing vessels. Jorg Wilting and his colleagues (Anatomical Institute II, Albert-Ludwigs-Universitat, Freiburg) have shown that VEGF is an important angiogenic factor in chicken embryogenesis: VEGF expression is coincident with regions of the embryo that become vascularized; and the VEGF receptor VEGFR-2/Flk-1 is an early marker of angioblast precursors of endothelial cells.

The importance of specific angiogenic factors and receptors in blood vessel formation during embryogenesis was reinforced by the results of gene knockout experiments. Homozygous deletions of the VEGF receptors VEGFR-1/Flt-1 and VEGFR-2/Flk-1 (Napoleone Ferrara, Genentech), the VEGF-C receptor Flt-4 and the receptor Tie-1 (Kari Alitalo, Haartman Institute, University of Helsinki) and FGF receptors 1 and 2 (Chuxia Deng, National Institute of Diabetes, Digestive and Kidney Diseases) were lethal to the developing embryo, as was the heterozygous deletion of the VEGF gene (N. Ferrara). The phenotypes of these lethal mutations included vascular defects ranging from the absence of blood vessels and blood cells in the VEGFR-2 null mutant to an incomplete vasculature and a nonfunctional placenta in the FGF receptor mutants. Thus, the ability to form a normal vasculature from multipotent stem cells is essential for the survival of the developing embryo, while the ability to utilize many of the same angiogenic factors to recruit blood vessel formation later in development is crucial to the survival of tumors undergoing malignant transformation.

A number of endothelial cell adhesion molecules such as VE-cadherin (Philippe Huber, Laboratoire de Transgenese et Differenciation Cellulaire, Grenoble), PECAM-1 (Joseph Madri, Department of Pathology, Yale University), fibronectin (Jean Schwarzbauer, Department of Molecular Biology, Princeton University), and the avb3 integrin and one of its ligands, Del1 (Thomas Quertermous, Vanderbilt University) are known or suspected to mediate the organization of endothelial cells or fetal cytotrophoblasts (Susan Fisher, University of California at San Francisco) into vascular structures during embryonic development. Many of these molecules are also likely to be involved in tumor vascularization; however, Del1, at least, is expressed only by embryonic endothelial cells.

Three talks described potentially unique aspects of the biology of the heparin-binding, angiogenic fibroblast growth factors 1 and 2 (FGF-1 and -2), the prototypic members of the 14 member fibroblast growth factor family. Both FGFs are potent endothelial cell mitogens in vitro and inducers of angiogenesis in vivo, and both growth factors confer a transformed phenotype to transfected murine fibroblasts. FGF-1 and -2 are secreted proteins that bind and activate four distinct cell surface receptor tyrosine kinases, yet neither is synthesized with a signal sequence thought to be needed for secretion. Moreover, forms of both molecules contain nuclear localization sequences. Thus, two longstanding enigmas in FGF research have been the mechanisms or mechanisms by which FGF-1 and -2 are released from cells to activate their receptors, and the functional significance of FGFs transported to the nucleus of the cell.

Tom Maciag (American Red Cross) and his colleagues have characterized in some detail a novel secretory pathway by which FGF-1 (acidic FGF) makes its way out of cells. The initial clues that led to this pathway's identification were the fortuitous findings that FGF-1 can form inactive dimers under appropriate conditions and that it is released by cells in response to heat shock. Further investigation showed that upon heat shock, FGF-1 is released as part of a protein complex consisting of an FGF-1 dimer, a proteolytic fragment of synaptotagmin, which is a protein component of vesicle membranes, and the calmodulin-binding protein S100A13. All three components of the secreted complex are phosphatidylserine (PS)-binding proteins, which suggests that the complex is translocated to the inner leaflet of the plasma membrane where it binds PS and is "flipped out" into the extracellular space. The latent FGF dimer is then activated by reduction. FGF-2 (basic FGF) does not dimerize, and it is not secreted from cells via the FGF-1 release pathway. Finally, although FGF-1 is released in response to heat shock, it is also released from cells in response to hypoxia, a condition that is commonly found in large solid tumors.

Dan Rifkin (Department of Cell Biology, New York University) and his colleagues have found that three N-terminally extended forms of FGF-2 can be synthesized from CUG initiation codons on FGF-2 mRNA. Unlike the 18 kDa form of FGF-2 synthesized from a single AUG codon, the high molecular weight (HMW) forms of FGF-2 are not secreted, but are instead translocated to the nucleus. This difference in destination is reflected by differences in biological effects: 18 kDa FGF-2 binds and downregulates surface receptor molecules; it stimulates integrin synthesis and increased cell motility; and it induces morphological changes in transfected NIH 3T3 cells. HMW FGF-2 has none of these effects. However, both HMW and 18 kDa FGF-2 confer the ability to grow in soft agar to 3T3 transfectants, which is a measure of transformation. Thus, it appears that 18 kDa FGF-2 transforms cells through an autocrine mechanism, while HMW FGF-2 transforms cells through an intracrine mechanism. This intracrine mechanism is not completely understood, but HMW forms of FGF-2 are methylated on 10-16 arginine residues in their N-terminal extensions, and these modified FGFs appear to concentrate in nucleoli where they bind RNA and perhaps nucleolar proteins. One might therefore speculate that nuclear FGF is involved in regulating ribosomal gene expression or general protein synthesis through an effect on ribosome assembly.

Wally McKeehan (Texas A&M University) and his colleagues have been investigating the role of heparan sulfate molecules in mediating the selectivity of FGF receptors for ligands. The four known FGF receptors bind FGFs with overlapping but distinct specificities. For example, the FGFR 2(IIIb) isoform expressed on epithelial cells binds FGF-1 and FGF-7 (keratinocyte growth factor) but not FGF-2 whereas FGFR-1 expressed on mesenchymal cells binds FGF-1 and FGF-2 but not FGF-7. The FGFs are known to bind heparan sulfate proteoglycans on the surface of cells, but the McKeehan lab finds that FGF receptors also bind heparan sulfate and that cell type-specific heparan sulfate molecules may dictate receptor ligand specificity. In addition, they find that the ectopic expression of FGFR-1 and the loss of FGFR-2 in prostate epithelial cells is associated with malignant progression. Whether or not FGFR-1 ligand preference is influenced by epithelial-specific heparan sulfate molecules is not known.

The highlight of the symposium for many attendees was the keynote address by Judah Folkman and the talk by his colleague Michael O'Reilly (Department of Surgery, Children's Hospital and Harvard Medical School). Dr. Folkman described his work as summarized above, focused on the hypothesis that tumor growth is angiogenesis-dependent, and provided evidence that leukemias are similarly dependent on or stimulated by angiogenic growth factors. Dr. Folkman also related how the observation that the resection of primary tumors sometimes results in the growth of metastases led his laboratory to discover tumor-derived inhibitors of angiogenesis. Two of these inhibitors, angiostatin and endostatin, were purified and characterized by Michael O'Reilly: angiostatin is a 30 kDa fragment of the zymogen plasminogen, and endostatin is an 18 kDa fragment of the capillary basement membrane component collagen XVIII.

In animal studies angiostatin and endostatin cause tumor regression by increasing the rate of tumor cell apoptosis. In the continued presence of either inhibitor, a dormant state is reached in which tumor cell proliferation is offset by apoptotic cell death. Tumors do not become resistant to these angiogenesis inhibitors, but they recur when inhibitor treatment is stopped. However, the Folkman lab finds that after a number of cycles of tumor growth and endostatin-induced regression, all experimental tumors thus far examined go into remission and do not recur.

The mechanism by which this unexpected curative effect is achieved is not known, but these impressive results suggest that anti-angiogenesis therapies may be used with more than tumor regression and dormancy as the therapeutic endpoints. Folkman offered second authorship on a paper to any symposium participant who could formulate a good hypothesis on how repeated cycles of endostatin treatment could lead to the eradication of tumors in vivo. This prize may have been claimed by Bob Auerbach for suggesting that cyclical treatment with endostatin induces a "Hayflick phenomenon" in which vascular endothelial cells reach a natural limit beyond which they can no longer proliferate, and the tumor dies. This writer, for one, will be searching MEDLINE in a year or so for a publication by O'Reilly, Auerbach, and Folkman.

Red Light for Metastatic Cancer - Harvard Medical School article profiling research on an endothelial-cell-specific angiogenesis inhibitor.

ANGIOGEN - an unmoderated mailing list designed mainly for Ph.D.-level scientists and medical doctors actively engaged in research on angiogenesis inhibitors and their possible clinical applications. Subscribe to it via the OncoLink Automated E-mail Discussion Group Subscriber.