Small Molecules in Large Proteins

A theme is emerging in antiangiogenesis research: Small molecules stored within large proteins in the body can stop cancer cells from creating new blood vessels. Many enzymes that a tumor uses to invade surrounding tissue generate these angiogenesis inhibitors, but a tumor can locally override the effect of the inhibitors by generating angiogenesis stimulators. If researchers could shift this balance by increasing the concentration of endogenous angiogenesis inhibitors, they could potentially ar

Nadia Halim
Aug 20, 2000

A theme is emerging in antiangiogenesis research: Small molecules stored within large proteins in the body can stop cancer cells from creating new blood vessels. Many enzymes that a tumor uses to invade surrounding tissue generate these angiogenesis inhibitors, but a tumor can locally override the effect of the inhibitors by generating angiogenesis stimulators. If researchers could shift this balance by increasing the concentration of endogenous angiogenesis inhibitors, they could potentially arrest tumor development. Without blood vessels, cells are unable to grow into tumors. Researchers have recently found a treasure chest of these molecules, which may bolster the clinical pipeline for cancer drugs.

Judah Folkman, director of the surgical research laboratory at Children's Hospital in Boston and professor of cell biology and pediatric surgery at Harvard Medical School, compares this facet of angiogenesis to the body's blood clotting mechanism. In both cases, proteins are made and stored in an inactive form so they can be produced quickly when needed; there may be no time for gene expression. For instance, normally blood clotting in the body must be inhibited, but fast and specific clotting is necessary in the case of an injury. Intricate control mechanisms in the body also have evolved to accommodate situations requiring a fast transition from long-term quiescence to a swift, localized angiogenic response.

Image: The Angiogenesis Foundation, Cambridge, Mass.

Many drugs targeting the various steps of tumor blood vessel formation are already in clinical trials. (AF= angiogenic factor, EC= endothelial cell, BM=basement membrane, MMPs=matrix metalloproteinases, Tie-2= receptor tyrosine kinase)
Noel Bouk, professor of microbiology at Northwestern University Medical School, discovered one of the first endogenous angiogenesis inhibitors, thrombospondin, a ubiquitous adhesive glycoprotein often found in blood clots and the extracellular matrix.1 The advantage to using endogenous molecules in therapy, says Bouck: "They are the molecules the body uses to keep the vasculature in check. You are not going to have antibodies or unusual reactions against them." Thrombospondin, however, is not being developed as a potential drug candidate because of its large size: 140 kDa. Bouck's lab also found that small peptides synthesized from this protein had significantly lower activity than the whole molecule had.

Angiostatin and endostatin, naturally occurring peptides derived from larger molecules, however, entered clinical trials last year. "Tumors make enzymes that cleave or change larger proteins present in the body. The smaller protein has a different function than the parent molecule," explains Folkman. For example, angiostatin, which inhibits angiogenesis, is derived from plasminogen, a clotting factor that normally circulates in the blood.2 Angiostatin, however, has nothing to do with clotting, and plasminogen cannot inhibit angiogenesis. Michael O'Reilly, postdoctoral fellow in Folkman's lab, has discovered two other angiogenesis inhibitors that are derived from larger proteins: endostatin from collagen XVIII3 and a cleaved form of antithrombin,4 which prevents thrombin from making unwanted clots in the blood.

Though angiostatin and endostatin function has been elucidated in the last few years, much work remains. The pathways and enzymes that generate angiostatin are known, and at least one receptor for it has been published. Salvatore Pizzo, professor of pathology at Duke University Medical Center, and colleagues identified ATP synthase as an angiostatin receptor on the surface of endothelial cells.5 Members of the Karolinska Institute in Stockholm have presented at meetings unpublished work on another receptor, indicates Folkman. No receptor has been found for endostatin, but the Folkman lab has published work that shows elastase can cleave collagen XVIII to form endostatin. "It's not uncommon to find molecules that work therapeutically and then discover their molecular mechanism of action. Aspirin and penicillin are examples," explains Folkman.

Vikas Sukhatme, professor of medicine at Harvard Medical School, discovered restin, another such molecule. Restin is derived from collagen XV, part of the vascular basement membrane, which is an insoluble structural wall that surrounds capillaries.6 Basement membranes are composed of extremely large proteins such as collagen, laminin, proteoglycans, fibronectin, and entactin. Restin and endostatin are composed of the noncollagenous domain (NC) of two different collagens. Collagen structure usually includes a triple helix and noncollagenous domains at the C-terminal that lack the triple helical structure.

Novel Family of Inhibitors

As researchers look into how existing angiogenesis inhibitors work, they are discovering new inhibitors. Two labs have independently identified a novel family of these molecules derived from collagen IV, another vascular basement membrane protein. Raghu Kalluri, assistant professor of medicine at Harvard Medical School, had studied the vascular basement membrane for many years but began appreciating its impact on new capillary growth more recently. "We wanted to understand what happens to all the fragments generated from the vascular basement membrane as it undergoes changes during angiogenesis. Our strategy was to beat up the membrane with proteases that are present in a tumor environment. Then we studied the resulting fragments to determine whether they were angiogenic or antiangiogenic," he says.

Collagen IV was a good first target for several reasons. Much of its biochemistry and molecular biology has been studied because of its abundance and its importance in the assembly of the basement membrane. Though endostatin and restin provided a promising precedent, collagens XVIII and XV, from which they are derived, are relatively rare compared with collagen IV. Therefore, less is known about them. Collagen IV is composed of six distinct gene products, *1-*6. After screening the antiangiogenic potential of all six NC1 domains, his lab found that three fragments had potent antiangiogenic activity. The researchers have named the NC1 domain of the *2 chain canstatin7; the NC1 domain of the *1 chain arresten8; and the NC1 domain of the *3 chain tumstatin.9

Canstatin and arresten inhibited proliferation, migration, and tube formation of the endothelial cells, in addition to disrupting basement membrane function. Canstatin also induced apoptosis without having an effect on nonendothelial cells and was more effective in causing apoptosis of endothelial cells under the influence of growth factors triggered by a tumor. In vivo testing showed that canstatin and arresten suppressed tumors and metastases in two human xenograft mouse models. The researchers suggest that arresten binds to *1ß1 integrin, which down-regulates vascular endothelial growth factor (VEGF)-induced proliferation and migration of endothelial cells. Integrins are a family of cell surface receptors that mediate adhesion of cells to extracellular proteins and sometimes to other cells.

Nicholas Kefalides, professor of medicine at the University of Pennsylvania School of Medicine, initially found that tumstatin inhibits melanoma cell proliferation. While the C-terminal peptide sequence for this activity was in amino acids 185-203, Kalluri's lab localized the antiangiogenic activity to amino acids 54-124. They went on to show that amino acids 54-124 bind to *vß3 integrin, which is not essential for inhibition of tumor cell proliferation but is necessary for antiangiogenic activity. A fragment of tumstatin containing the 185-203 amino acid sequence binds both endothelial and melanoma cells but only inhibits the melanoma cell proliferation. The challenge will be to administer a drug with both activities; it may require a protein engineered in a way that exposes both motifs.

Kalluri's lab is evaluating which molecules will be better for clinical development. The National Cancer Institute is doing independent tests too. "The mechanism of action is different for all three molecules, which is exciting. They are binding to different cell surface receptors and therefore activating different pathways," says Kalluri. This may mean a tremendous potential for combination therapy, he adds.

Peter Brooks, assistant professor of biochemistry and molecular biology at the University of Southern California School of Medicine, has also screened the six NC1 domains of collagen IV in collaboration with Billy Hudson and BioStratum Inc., Research Triangle Park, N.C. Hudson and Michael Sarras, professors of biochemistry and molecular biology at University of Kansas Medical Center, showed that isolated NC1 domains could inhibit hydra differentiation. Then Biostratum, a company looking at the basement membrane as a source of therapeutic molecules, obtained preliminary results indicating antiangiogenic activity of certain NC1 domains.

The Brooks lab found that the NC1 domains of *2, *3, and *6 inhibit angiogenesis by binding to specific integrins in an RGD-independent manner.10 The RGD (arginine, glycine, aspartic acid) amino acid tripeptide is believed to be a binding site for a specific subset of integrins, particularly *vß3. However, the *2 and *6 NC1 domains seem to bind *vß3 integrin and support cell adhesion independently of RGD. "We were surprised by the finding," says Brooks. Kalluri's lab also confirmed this for the *3 NC1 domain. The Brooks lab has looked at the molecules in the chick embryo and a number of different in vivo assays and now is looking at them in the human/mouse chimeric model. In this model they can look at human angiogenesis within a patch of human skin on a mouse. "We have a number of papers under review right now," he says.

The NC1 domains could be acting in a number of different ways. Because the molecules directly mediate interactions with endothelial cells through integrins, the soluble form of the NC1 domains may be disrupting integrin-dependent interactions, preventing migration and adhesion, for example. Some data suggest that they may be blocking integrin-dependent signaling for endothelial cell proliferation. Another interesting mechanism, according to Brooks and Hudson, involves the NC1 domain's function in matrix assembly. During the early phases the matrix has to be disassembled or proteolyzed, and during the late phases it has to be reestablished. These isolated NC1 domains might be preventing the structural assembly of the matrix and thereby inhibiting angiogenesis.

Targeting integrins on the endothelial cells circumvents problems associated with targeting angiogenic factors secreted by tumors, such as VEGF or fibroblast growth factor (FGF). The tumor may produce different factors after a time period or after metastases so that an inhibitor against a specific factor would no longer be effective. "There is a plethora of different cytokines. If you block one, there's more that may do similar things," explains Brooks.

Kalluri and Brooks caution that these molecules have just been discovered, so it will be a couple of years before clinical trials. Endostatin, discovered in 1996, just entered clinical trials last year. However, the researchers point out that the biology and the mechanism of action are better understood with these collagen IV molecules than with endostatin at its early stage of development. "Our initial papers are talking about how these molecules work and their molecular mechanism of action, so in comparison to other molecules, we know quite a bit in a very short period of time. Hopefully this will lead to a faster path to clinical trials," comments Kalluri.


Advantages and Disadvantages

"There are many ways to stop angiogenesis, and I don't think they are mutually exclusive," declares Folkman. However, one powerful advantage of endogenous angiogenesis inhibitors is the lack of side effects. Endostatin, by evolutionary evidence, has been around for at least 600 million years. Such highly conserved molecules are not usually toxic. In fact, patients who have been on endostatin as long as eight and a half months do not exhibit any side effects. This brings up another advantage: lack of drug resistance. The target of antiangiogenesis therapy is normal endothelial cells that line the blood vessel, not the unstable cancer cell, which can mutate and acquire drug resistance. Endothelial cells, on the other hand, divide normally without mutation and do not develop resistance. Another advantage is high specificity. Endostatin turns off new blood vessel growth--mainly tumor blood vessel growth. Synthetic inhibitors may lack such specificity. For instance, TNP-470, which is a synthetic form of a fungal protein, slows wound healing and stops pregnancy in mice.

However, there are disadvantages to protein therapy. It must be taken intravenously. It is also expensive, because companies do not have much experience at making large quantities of proteins. Pharmaceutical companies have become adept at developing orally bioavailable, small molecules, but only a few companies are efficient at making proteins, says Folkman.

Gerald Soff, assistant professor of hematology and oncology at Northwestern Medical School, has taken an interesting approach to the problem of protein therapy. He developed a way for the body to make its own angiostatin. When given to a patient in combination, two approved drugs, captopril (Bristol-Myers Squibb Co., Princeton, N.J.) and tissue plasminogen activator, allow the body to make angiostatin, which is active in vivo and suppresses metastases. "If you had all the angiostatin you wanted you probably wouldn't do it this way, but in the meantime it's a good demonstration of principle," explains Folkman.

Another option, gene therapy, is coming along mainly in the laboratory and not in people. Folkman mentions that Novartis is one of the leaders in this area, and it presented a Gordon Conference poster about a year ago demonstrating endostatin delivery through gene therapy. "It would be the ultimate way to do it because you could get an injection every five months for adenoassociated vectors, for example," explains Folkman.


More Inhibitors to Come

"For a long time people have been looking at growth factors and their receptors, proteases, and cell surface receptors. But now people are seeing the matrix outside the cell as a vast wealth of information that can control angiogenesis," says Brooks. Once thought to be simply a mechanical scaffold, vascular basement proteins are now considered potential therapeutic targets. By focusing on the interaction of endothelial and tumor cells with the extracellular matrix, researchers may be able to block angiogenesis, tumor invasion, and metastasis at the same time. "There is a growing push to move this new approach forward. In fact we are seeing a lot more of it in terms of small start-up companies popping up," he adds.

Kalluri emphasizes understanding how the vascular basement membrane regulates angiogenesis, both the initiation and the braking. Domains in the basement membrane, which may be inherently antiangiogenic, may keep endothelial cells in a resting state. Then when tumor cells provide angiogenic cues such as growth factors, the basement membrane breaks down in preparation for endothelial cell migration and new blood vessel formation. "If we understand this process, then we may understand why these inhibitors only affect blood vessels in a tumor environment and not preestablished vasculature," Kalluri says. Researchers have a handful of endogenous antiangiogenesis molecules so far, he adds, "but there may be a bag of 20 or 30 inhibitors that work together to give an effective concentration and provide the brakes." S


Nadia S. Halim can be contacted at



1. R. Rastinejad et al., "Regulation of the activity of a new inhibitor of angiogenesis by a cancer suppressor gene," Cell, 56:345-55, 1989.

2. M.S. O'Reilly et al., "Angiostatin: a novel angiogenesis inhibitor that mediates the suppression of metastases by a Lewis lung carcinoma," Cell, 79:315-28, 1994.

3. M.S. O'Reilly et al., "Endostatin: an endogenous inhibitor of angiogenesis and tumor growth," Cell, 88:277-85, 1997.

4. M.S. O'Reilly et al., "Antiangiogenic activity of the cleaved conformation of the serpin antithrombin," Science, 285:1926-8, 1999.

5. T.L. Moser et al., "Angiostatin binds ATP synthase on the surface of human endothelial cells," Proceedings of the National Academy of Science, 96:2811-6, 1999.

6. R. Ramchandran et al., "Antiangiogenic activity of restin, NC10 domain of human collagen XV: comparison to endostatin," Biochemical and Biophysical Research Communication, 255:735-9, 1999.

7. G.D. Kamphaus et al., "Canstatin, a novel matrix-derived inhibitor of angiogenesis and tumor growth," Journal of Biological Chemistry, 275:1209-15, Jan. 14, 2000.

8. P.C. Colorado et al., "Anti-angiogenic cues from vascular basement membrane collagen," Cancer Research, 60:2520-6, May 1, 2000.

9. Y. Maeshima et al., "Two RGD independent *vß3 integrin binding sites on tumstatin regulate distinct anti-tumor properties," Journal of Biological Chemistry, 275: 23745-50, Aug. 4, 2000.

10. E. Petitclerc et al., "New functions for non-collagenous domains of human collagen," Journal of Biological Chemistry, 275:8051-61, March 17, 2000.