Cancer and Metastasis Reviews 19: 19–27, 2000.

© 2000 Kluwer Academic Publishers. Printed in the Netherlands.

The contributions of cyclooxygenase-2 to tumor angiogenesis

Stephen Gately
Division of Hematology/Oncology, Department of Medicine, Northwestern University Medical School,
Chicago, IL, USA

Key words: angiogenesis, apoptosis, cyclooxygenase-2, prostaglandins, vascular endothelial growth factor

Abstract

Cyclooxygenase-2 (COX-2) is an immediate early response gene that can be induced by a variety of tumor promoters,
cytokines, growth factors and hypoxia. COX-2 overexpression is linked to all stages of carcinogenesis with the
enzyme localized to the neoplastic cells, microvascular endothelial cells, and stromal fibroblasts. The contributions
of COX-2 in tumor angiogenesis include: (a) the increased expression of the proangiogenic growth factor VEGF;
(b) the production of the eicosanoid products thromboxane A2 , PGE2 and PGI2 that can directly stimulate endothelial
cell migration and growth factor-induced angiogenesis; and potentially, (c) the inhibition of endothelial cell apoptosis
by stimulation of Bcl-2 or Akt activation. Selective pharmacological inhibitors of COX-2 as angiosuppressive agents
could have therapeutic benefit in the treatment of neoplastic disease from prevention through treatment of advanced
metastatic disease. These agents are safe and well tolerated and can be added to chemotherapy and radiation therapy
where angiogenesis inhibitors appear to provide at least additive therapeutic benefit.

Classical therapeutic strategies for neoplastic diseases
focused on the cancer cell. These approaches achieved
only limited success with nearly half of all diagnosed
cancer patients dying as a result of their disease despite
aggressive treatment with highly toxic regimens [1].
Because tumor growth and metastases are depen-
dent upon angiogenesis [2–5], therapeutic strategies
that target the microvasculature are being evaluated
clinically [6].
Tumor angiogenesis results from a cascade of molec-
ular and cellular events [4], often initiated by the release
of angiogenic growth factors [7,8]. Tumor cells express
a variety of proangiogenic growth factors that diffuse
in the direction of pre-existing blood vessels [2,8,9].
These growth factors activate the normally quiescent
vascular cells inducing: (i) proteolytic degradation
of the basement membrane [10,11]; (ii) migration
of endothelial cells towards the angiogenic stimulus
[12,13]; (iii) endothelial cell proliferation [12,14,15];
(iv) lumen formation [10,12,13,16]; (v) pericyte cap-
ping [12,17–19]; and (vi) production of a new base-
ment membrane. Tumors can also acquire vasculature
through vasculogenesis, the formation of blood ves-
sels from progenitor endothelial cells, or angioblasts
[20–22]. Vasculogenesis was previously thought to be
restricted to the embryo [23,24], however, circulating
stem cells in peripheral blood have been detected that
can differentiate into endothelial cells and contribute
to angiogenesis in the adult [25–27].
Angiogenesis and cancer
It is generally accepted that solid tumor growth and
metastases are dependent upon the acquisition of
an adequate blood supply [2,4,28–31]. Pharmaco-
logical targeting of the microvasculature in patients
with neoplasms represents an attractive therapeutic
approach because the inhibition of angiogenesis has
been shown to prevent growth [32], and can induce
regression of experimental solid tumors [33]. Further-
more, antiangiogenic therapy could be utilized as adju-
vant treatment for microscopic metastases, preventing
the emergence of a vascular supply and maintaining
tumor dormancy [34,35]. Because of the normally slow
renewal of vascular stroma, even in tissues with high
epithelial turnover such as skin and jejunum, antiangio-
genic therapies promise to be free of the hematopoetic,
20
gastrointestinal and other toxicities of standard antipro-
liferative therapies [33,36], and appear not to induce
acquired drug resistance [33]. Recent data suggest that
selective inhibitors of cyclooxygenase-2 (COX-2) are
potent inhibitors of angiogenesis [37,38] suggesting
the potential utility of these agents in oncology. The
remainder of this review will focus on the contribution
of COX-2 in tumor angiogenesis.
COX-1 and COX-2
COX-1 and COX-2 are the two enzymes that convert
arachidonic acid to prostaglandins [39–41]. COX-1
is constitutively expressed and is responsible for
normal kidney and platelet function as well as the
maintenance of the gastrointestinal mucosa [41]. In
contrast the COX-2 enzyme can be induced by a vari-
ety of proinflammatory cytokines such as IL-1 [42],
growth factors such as epidermal growth factor (EGF)
[43], transforming growth factor-beta (TGF-b) [44,45],
tumor necrosis factor-alpha (TNF-α) [45,46], inducible
nitric oxide synthase (iNOS) and ultraviolet B light
(UVB) [40]. Indomethacin, a non-selective inhibitor of
cyclooxygenases has been shown to inhibit angiogen-
esis in experimental models of angiogenesis induced
by tumors [47], thermal burns [48], copper salts [49]
or prostaglandins [50]. Indomethacin also suppressed
lymphocyte-mediated angiogenesis in vascularized
fascia [51]. More recently, indomethacin was shown
to inhibit angiogenesis alone [52], and when com-
bined with interleukin-12 [53]. Because non-selective
cyclooxygenase inhibitors are often associated with
renal and gastrointestinal toxicities, selective inhibitors
of the COX-2 enzyme have been developed that may
be ideal in chronic administration for angiogenesis
inhibition in the oncology setting.
COX-2
The significant contribution of COX-2 in cancer pro-
motion was elegantly demonstrated experimentally in
a model of human familial adenomatous polyposis,
where mice, genetically predisposed for polyp for-
mation by a targeted truncation deletion in the APC
tumor suppressor gene, were crossed with COX-2−/−
mice [54]. The tumor burden of the double mutant
offspring was significantly reduced by the genetic
knockout of COX-2, furthermore the reduction in
polyp formation was equal to that achieved by treat-
ing the polyp forming mice with a selective COX-2
inhibitor [54]. The pharmacologic application of selec-
tive COX-2 inhibitors also has been demonstrated to
significantly reduce the growth of a variety of experi-
mental tumors including head and neck [55], skin [56],
colon [57] and bladder [58].
Cellular expression of COX-2
The elevated expression of COX-2 in the cancer setting
has been localized to the neoplastic epithelium [59],
within the microvasculature [37] and stromal cells [60].
Tumor associated COX-2 expression. The potential
importance of tumor-associated COX-2 expression for
angiogenesis and tumor growth was demonstrated
when COX-2 expressing tumor cells were found to
grow larger and were more angiogenic in comparison
to implanted tumor cells that lacked COX-2 expres-
sion [59]. Additionally, tumor growth and angiogenesis
could be suppressed by selective COX-2 inhibitors only
if the tumor cells expressed COX-2 [61]. Together
these data demonstrate that tumor-associated COX-2 is
required for angiogenesis and tumor growth, and sug-
gest that the potential effectiveness of selective COX-2
inhibitors is determined by expression of COX-2 by
the tumor cell. One mechanism for the elevated tumor-
associated COX-2 expression could be related to the
marked repression of transcription of the COX-2 gene
by wild-type p53 [62]. Loss of wild-type p53 would
then be associated with the increased expression of
COX-2 mRNA and protein [62], increased secretion
of VEGF and decreased production of an endoge-
nous inhibitor of angiogenesis, thrombospondin-1
[63,64] resulting in the acquisition of a proangiogenic
phenotype.
Microvascular COX-2 expression. Previous studies
suggested a role for endothelial cell-derived COX-1 in
angiogenesis [59,61], however, a recent study demon-
strated that selective COX-2 inhibitors could suppress
growth factor-induced angiogenesis, and suggested
that endothelial-derived COX-2 is essential in directly
regulating angiogenesis [52]. A recent histochemical
evaluation of human colon, breast, prostate and lung
tumor biopsies demonstrated that COX-1 immunore-
activity is broadly distributed within normal and neo-
plastic tissues, whereas COX-2 immunoreactivity is