| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
,1
* Department of Oncology, Georgetown University Medical Center, Washington, D.C. 20007, USA;
Human Genome Sciences, Rockville, Maryland 20850, USA;
Department of Radiology, Rochester University Medical Center, Rochester, New York 14642, USA;
Department of Biology, Xiamen University, China;
¶ Laboratory of Molecular Biology (Celgen) and Department of Cell Growth, Differentiation and Development, Flanders Interuniversity Institute of Biotechnology (VIB), University of Leuven, Leuven, Belgium; and
** Laboratory of Molecular Biotechnology, Department of Biochemistry, Universiteitsplein 1, Wilrijk, Belgium
3Correspondence: Department of Oncology, Georgetown University Medical Center, 3970 Reservoir Road, NW, Washington, D.C. 20007, USA. E-mail: zhangl{at}georgetown.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: angiogenesis • breast cancer • ECM1 • endothelial cells
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Before the formation of blood
vessels, most tumors are relatively small, remain localized, and grow
slowly. However, with the advent of vascularization, the tumors become
more malignant, as characterized by rapid growth, invasiveness, and
metastasis. In many cases, tumor cells facilitate their own progression
by directly producing angiogenic factors such as vascular endothelium
growth factor (VEGF) and fibroblast growth factors (FGFs), to stimulate
a persistent angiogenesis that leads to their uncontrolled growth and
metastasis (1
2
3)
. Alternatively, some tumors act by
inducing the adjacent normal cells to synthesize the angiogenic factors
(4)
. In either case, the process of neovascularization is
a common feature of malignant progression of solid tumors and as such
represents a potential target for inhibition of tumor growth and
metastasis.
In this study, we examined the angiogenic properties of extracellular
matrix protein 1 (designated ECM1 in the human and Ecm1 in the mouse)
that was initially isolated from an osteogenic stromal cell line
(5)
. Although earlier sequencing studies indicated that
ECM1 was not directly homologous to any other known proteins, it did
contain six cysteine doublets having a
CC-(X7–10)-C arrangement, similar to those in
serum albumin (6
7
8)
. Such sequences can generate
"double-loop" domains that are likely involved in ligand binding
(9
, 10)
. In humans, the ECM1 message is differentially
spliced, giving rise to two forms of the protein: a long
form (ECM1a) composed of 540 amino acids, and a short form (ECM1b) of
415 amino acids that lacks the amino acids derived from the central
seventh exon (7)
. Northern analysis of different human
tissues has indicated that the 1.8 kb transcript for ECM1a was
expressed predominantly in the blood vessel-rich placenta and heart,
whereas the 1.4 kb mRNA for ECM1b was present in the tonsils and skin
(6
, 7
, 11)
.
Previous studies have suggested that mouse Ecm1a plays a regulatory
role in the process of endochondral bone formation (12)
.
When recombinant human ECM1a was added to organ cultures of metatarsals
isolated from mouse embryos, it inhibited both alkaline phosphatase
activity and mineralization in a dose-dependent fashion. In the mouse
embryo, Ecm1a mRNA was expressed by perichondral connective tissue but
not by the chondrocytes. Thus, Ecm1a produced by perichondral tissue
appears to act in a paracrine fashion and can be considered a negative
regulator of endochondral bone formation. However, beyond this, little
was known about other possible functions of human ECM1.
Here, we report that ECM1 also possesses angiogenic-promoting properties as indicated by its ability to stimulate the formation of blood vessels in the chorioallantoic membrane (CAM) of chicken eggs. The angiogenic property of ECM1 was also suggested by its close association with embryonic endothelial cells and its ability to stimulate the proliferation of endothelial cells in tissue culture. Furthermore, the fact that ECM1 was up-regulated in some breast cancer cells suggests that it may play a role in tumor progression.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
.
Purification of human ECM1
The human cDNA sequences encoding ECM1a and ECM1b proteins were
amplified by polymerase chain reaction using a human fetal liver cDNA
library as a template and the following primers: 5'-primer:
5'-CGGGATCCGCCATCATGGGGACCACAGCCAG-3', consisting of a
BamHI restriction site, a "Kozak" sequence, and the
first 17 bases of the open reading frame; 3'-primer:
5'-GCTCTAGATCCAAGAGGTGTTTAGTG-3', containing an XbaI
restriction site followed by 18 nucleotides complementary to the
3'-untranslated sequence. The amplified fragments were cloned into the
baculovirus expression vector pA2. The generation of recombinant
baculoviruses and expression of ECM1 were performed as described
previously (14
, 15)
.
For isolation of the ECM1a protein, conditioned medium from the virus-infected insect cells was directly applied to a strong cation exchange column (Poros HS, PerSeptive Biosystems, Framingham, Mass.) equilibrated with 0.05 M NaCl, 0.02 M Bis-Tris, pH 6.0, and 10% (v/v) glycerol (buffer A). The protein was eluted with a step gradient of a high ionic strength buffer consisting of 35% 1.0 M NaCl, 0.02 M Bis-Tris, pH 6.0, and 10% (v/v) glycerol (buffer B). The fractions containing protein were pooled and diluted with water to reduce the ionic strength. The sample was clarified by centrifugation and was then applied to a strong anion exchange column (Poros HQ) connected in tandem to a strong cation exchange column (Poros HS) pre-equilibrated with buffer A. The columns were eluted with a gradient from 0 to 70% buffer B. The fractions containing ECM1a, as determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), were pooled, diluted with water, and adjusted to pH 8.0 with 0.5 M Bis-Tris propane (pH 9.0). The resulting solution was then applied to a Poros HQ column equilibrated with buffer A at pH 8.0. The proteins were eluted using a gradient of from 0 to 100% buffer B at pH 8.0.
ECM1b was purified in a similar fashion with only slight modification of the buffers. For the first cation exchange column (Poros HS), the proteins were eluted with a step gradient of 35% 2.0 M NaCl, 0.02 M Bis-Tris, pH 6.0, and 10% (v/v) glycerol (buffer C). The ECM1b-containing peaks were pooled, diluted, and adjusted to pH 8.0. For the second columns (Poros HS and HQ connected in tandem), the proteins were eluted with a gradient of from 0 to 50% 2.0 M NaCl with 0.02 M Tris pH 8.0. For the final purification step, the fractions containing ECM1b, were pooled, diluted, and adjusted to pH 5.0; applied to an HS column; and eluted with a gradient of from 0 to 100% buffer C.
The fractions containing purified ECM1a or ECM1b were analyzed by
SDS-PAGE, and those containing a single band were pooled. The purified
proteins were stored as a stock solution of 
1.4 mg/ml in
phosphate-buffered saline, pH 7.3. The identity of these preparations
was verified by N-terminal amino acid sequencing, and the purity was
greater than 98% as determined by reverse-phase high-pressure liquid
chromatography. Endotoxin levels of the preparations were less than 5
EU/mg protein as determined by the Amebocyte Lysate Test
(Bio-Whittaker, Walkersville, Md.).
Antibodies
Antiserum against human ECM1 was raised by injecting 0.2 mg of
highly purified recombinant ECM1a or ECM1b in Freund’s complete
adjuvant (Difco Laboratories, Sparks, Md.) subcutaneously into rabbits.
The injections were repeated after 3 weeks and the rabbits were bled
every third week. Although the antisera generated by immunization with
ECM1a cross-reacted with ECM1b and vice versa, they were specific for
ECM1 as determined by ELISA and immunoelectrophoresis with the
recombinant proteins. In some cases, the antibodies were further
purified by affinity chromatography on ECM1 coupled to Sepharose 4B.
Cell culture
The human breast cancer cell lines (MCF-7, Hs578T, MDA-435,
MDA-436, MDA-468, Sk-Br-3, ZR571, and T47D), TSU prostate cancer cells,
and NIH-3T3 cells were purchased from the American Type Culture
Collection (Rockville, Md.). The LCC-15 (derived from a bone metastasis
of a breast cancer patient), the human umbilical vein endothelial cells
(HUVEC), and the primary as well as metastatic breast cancer tissues
were obtained from the Tumor Bank of the Lombardi Cancer Center,
Georgetown University, Washington, D.C. Adult bovine aorta endothelial
cells (ABAE) were kindly provided by Dr. Luyuan Li (Lombardi Cancer
Center), and the bovine retinal endothelial cells (BREC) and bovine
capillary endothelial cells (BCE) were the gift of Dr. Higger
(Department of Pediatrics, Georgetown University Hospital). The tumor
cells were maintained in 5% fetal bovine serum (FBS) plus 95%
Dulbecco’s modified Eagle’s medium (DMEM), and the endothelial cells,
in 10–20% FBS plus 90–80% DMEM containing 10 ng/ml basic FGF.
Proliferation assays
In the case of the endothelial cell lines (passage 4 of HUVEC,
BCE, and ABAE), the cells were subcultured into 96-well plates (2 x 103 cells/well) in 100 µl of endothelial
cell culture media (20% FBS-DMEM containing 10 ng/ml of FGF-2 and
VEGF). After 12 h, the medium was replaced with serum-free DMEM,
and after 24 h the indicated amounts of ECM1a, heat-inactivated
ECM1a (placed in a boiling water bath for 30 min), FGF-2, and VEGF were
added. After 24 h, [3H]TdR (0.3
µCi/well) was added to the medium, and the next day the cells were
harvested and the amount of incorporated
[3H]TdR was determined with a ß-counter. In
the case of the other cell lines (MDA-435, MDA-468, TSU, and NIH-3T3),
the protocol was the same as above except that ECM1 was added in medium
containing 1% FBS plus 99% DMEM. In each case, all experiments were
done in triplicate.
Histochemical staining
Cultured cells were grown overnight in 8-well chamber slides and
then fixed with 3.7% formalin for 5 min. The staining of these cells
as well as sections of human breast cancers was carried out by using
the same procedure as follows. The samples were incubated with rabbit
anti-ECM1b (1:400), followed by biotinylated goat anti-rabbit (1:250),
peroxidase-labeled streptavidin, and the peroxidase substrate
3-amino-9-ethyl-carbazole, and
H2O2, which gives a dark
red reaction product. The cells were counterstained with hematoxylin
(blue) and preserved with CrystalMount (Biomeda, Foster City, Calif.).
Western blotting
Cell lysates containing 30 µg of protein from different breast
cancer cell lines were subjected to electrophoresis on a 10%
SDS-polyacrylamide gel, transferred to a nitrocellulose membrane, and
were incubated with rabbit anti-ECM1b (1:1000) and peroxidase labeled
anti-rabbit antibodies (1:4,000) followed by enhanced
chemiluminescence.
Angiogenesis assay
This assay was performed according to the methods of Brooks et
al. with some modifications (16)
. The top air sac portions
of 7-day-old chicken eggs were opened and the CAMs exposed. Two days
later, filter disks (Whatman No. 1, 0.5 cm diameter) containing varying
concentrations of ECM1 and VEGF (15 µl of 0, 0.5, 1, and 2 µg/ml
solutions) were applied to the CAMs. Care was taken to place the
filters on regions of the CAM that were relatively deficient in
preexisting blood vessels. Each group consisted of 10 eggs. Additional
ECM1 and VEGF were applied to the filter discs for each of 3
consecutive days. On day 4, the CAMs and associated discs were cut out
and immersed immediately in 3.7% formaldehyde. For computer analysis,
the discs were divided into four quarters with fine wires. The blood
vessels in each quarter of the CAM were digitally photographed, and the
results were analyzed by use of an Optimas 5 program to
calculate the vessel area and length in the viewing area. The total
length and area of the vessels in the four quadrants of each disc were
normalized to the total area measured and expressed as vessel length or
area index. The means and the standard errors were calculated from all
quadrants of all discs in each group, and the statistical significance
was examined by Student’s t test.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, a strong Ecm1 signal was detected in almost all of the newly formed
blood vessels. The mRNA for Ecm1 was particularly prominent in the
endothelial cells lining the heart, aortic arch, pulmonary trunk,
thoracic aorta, and dorsal aorta (Fig. 1A
and E
).
In the brain, the signal formed a striated pattern, indicative of new
blood vessels (Fig. 1C
). Additional studies indicated that
this pattern of Ecm1 expression occurred only during specific stages of
embryonic development. In older embryos (embryonic day 15.5), the
signal dramatically decreased in the central nervous system and
completely disappeared from the walls of arteries and the heart (data
not shown). Thus, the expression of Ecm1 in blood vessels occurred
during a specific window of embryonic development in the mouse.
|
As shown in Fig. 1G
and H
, the expression pattern
of Ecm1 in the central nervous system was very similar to that of the
VEGF receptor flk-1, a marker of endothelial cells (17)
.
The most striking difference between flk-1 and Ecm1 was in the timing
of their expression. As previously reported, flk-1 mRNA persists
throughout postnatal development (17)
. In contrast, Ecm1
mRNA was dramatically down-regulated during later stages of gestation,
coinciding with the establishment of the blood-brain barrier on
embryonic day 14.5 (18)
. These results suggested that Ecm1
was closely associated with early stages of angiogenesis and prompted
us to investigate its role in this process.
ECM1 stimulated endothelial cell proliferation
Next, we examined the effect of ECM1 on the proliferation of
cultured endothelial cells. For this, varying concentrations of
recombinant ECM1a were added to the cultures of HUVEC in serum-free
medium, and the extent of proliferation was determined by the
incorporation of radioactive thymidine. As shown in Fig. 2A
, ECM1a significantly stimulated the growth of these cells,
with a maximal response at 20 ng/ml. We then compared the
stimulatory effects of ECM1a with those of both FGF-2 and VEGF, known
stimulators of endothelial proliferation. As shown in Fig. 2B
, the effects of ECM1a were significant but not as great
as those of either FGF-2 or VEGF (in each case the optimal
concentration was used). In addition, heat inactivation of the ECM1
preparation abolished the effects of ECM1a (Fig. 2B
).
Similar results were obtained with ECM1b (data not shown). ECM1 (both a
and b forms) also stimulated the proliferation of other endothelial
cell lines such as ABAE, BREC, and BCE (data not shown). In contrast,
when ECM1a over a range of concentrations was added to cultures of
MDA-435 and MDA-468 cells, no obvious effect on thymidine incorporation
was apparent (Fig. 2C
and D
, respectively). There
was a similar lack of effect on other nonendothelial cell lines tested
(TSU and NIH-3T3). Thus, the biological activity of ECM1a appears to be
at least partially selective in that it stimulates the proliferation of
endothelial cells but not other cell types that we tested.
|
ECM1 promoted angiogenesis
To determine whether ECM1 acts as an angiogenic factor in
vivo, we examined its effect in the chick CAM assay. Recombinant
ECM1a and ECM1b were adsorbed onto small pieces of filter paper that
were then applied to the CAMs of chicken eggs. Two days later, the CAMs
in direct contact with the filters were photographed and processed for
image analysis. Figure 3B
, C
, D
, E
shows that ECM1a stimulated
angiogenesis in a dose-dependent fashion and that a similar effect was
obtained with ECM1b. The effects were comparable to that of VEGF, which
was used as positive control (Fig. 3F
). Image analysis of
these tissues revealed that both the vessel length and area indices
(Fig. 3G
and H
) were stimulated by application of
ECM1a and ECM1b. These results suggest that ECM1 promotes angiogenesis
either directly by stimulating endothelial cells or indirectly by
induction of other factors.
|
ECM1 was expressed by human breast cancer cells
In many cases, tumor cells have been shown to secrete angiogenic
factors that are associated with malignant progression
(1
2
3)
. Furthermore, some tumor cells express proteins
that are normally present only during embryonic development
(19
20
21)
. Consequently, we decided to determine whether
tumor cells also express ECM1 by Western blotting using antibodies
specifically directed against ECM1. With such antibodies, we tested a
panel of human breast cancer cell lines (Hs578T, MDA-468, MDA-435,
Sk-Br-3, ZR571, T47D, and MCF-7) and found that only the MDA-435 cell
line expressed high levels of ECM1 protein (Fig. 4
). On the basis of its approximate size of 68 kDa, we believe that this
protein corresponds to the ECM1a isoform, because purified recombinant
ECM1a from CHO cells yielded a protein of similar molecular size (data
not shown). Histochemical staining of cultured MDA-435 cells indicated
that most of the ECM1 was located focally in the cytoplasm, where it
was probably associated with organelles of the secretory pathway
(Fig. 5D
). Interestingly, in our hands, the MDA-435 cells were also
the most malignant cell line, as judged by their growth in nude mice
(unpublished observations). Subsequently, we found that other cell
lines, such as LCC15 that was derived from a bone metastasis of breast
cancer, also express ECM1 (22)
. The expression of ECM1 in
malignant tumor cells suggests that it may play a role in tumor
progression.
|
|
To determine whether expression of ECM1 was associated with breast
tumors, we used the antibody to stain a panel of sections of both
primary and secondary tumors from human cancer patients. Figure 6
shows examples in which ECM1 was detected in primary breast cancer
cells (Fig. 6A
and B
) and secondary tumors formed
in the lymph nodes (Fig. 6C
) as well as in the bone marrow
(Fig. 6D
). In contrast, little or no staining was apparent
in the normal breast ductal cells, stromal fibroblasts, and
inflammatory cells (Fig. 6
small arrows). Of the definitive breast
cancer biopsy samples that we tested, 68% (21 of 31) were positive for
ECM1. Of these, 12 of 18 cases were metastatic tumors and 9 of 13 cases
were primary tumors. Thus, expression of ECM1 is frequently associated
with breast cancer cells.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 3)
.
However, the biological activity of ECM1 also extends to at least some
other cell types, such as chondrocytes for which ECM1a has been shown
to inhibit alkaline phosphatase activity and mineralization
(12)
. To our knowledge, this is the first report
suggesting that ECM1 plays a role in angiogenesis.
Our in situ hybridization studies have indicated that Ecm1
mRNA is closely associated with blood vessels at different stages of
mouse embryogenesis, particularly during midgestation. The signal is
present in newly formed vessels, and its distribution is very similar
to that of flk-1 (one of the receptors for VEGF), which has been
identified as a marker for endothelium (17)
. However,
unlike flk-1, Ecm1 mRNA expression was down-regulated before birth. We
propose that Ecm1 is a novel marker for embryonic angiogenesis. This
could also account for the fact that high levels of ECM1 are present in
the placenta, which contains numerous newly formed blood vessels.
Another organ system in which Ecm1 may be active is the skin. The mRNA
of ECM1 is associated with the human epidermis but not the underlying
dermis (11)
. It is tempting to speculate that epidermal
keratinocytes secrete ECM1 into the adjacent dermis and stimulate the
formation of blood vessels. Functionally, this area requires a high
degree of vascularization so that the avascular epidermis can receive
oxygen and other nutrients from underlying dermis for their
proliferation.
ECM1 may also play a role in tumor angiogenesis, which is a
prerequisite for rapid tumor growth and metastasis (1
2
3)
.
In many cases, tumor cells facilitate their own progression by directly
producing angiogenic factors or by inducing other cells to synthesize
them (1
2
3
4)
. ECM1 may represent a new factor among those
produced by tumor cells (such as MDA-435 and LCC-15) to promote their
progression. In our hands, the MDA-435 line was also the most
aggressive with respect to growth in nude mice (unpublished
observations). More important, when both primary and secondary breast
cancer tissues were tested for ECM1, a relatively high proportion of
the tumor samples was positive. In contrast, we found little or no
positive staining in normal breast ductal epithelial cells,
fibroblasts, leukocytes, and other stromal cells. Thus, it appears that
at least a proportion of tumor cells produce ECM1, which could
stimulate vascularization and promote tumor cell progression.
In addition to stimulating the proliferation of endothelial cells, ECM1
proteins may have a number of other effects on these cells. This is
suggested by the fact that during angiogenesis, endothelial cells must
carry out a number of processes, including 1) the removal of the old
basal lamina, 2) cell division, 3) migration to a new location, and 4)
the synthesis of a new basal lamina (23)
. Indeed, other
angiogenic factors such as FGFs and VEGF have been found to have
multiple effects on endothelial cells with regard to proliferation,
migration, and production of proteases (1)
. Thus, it would
be interesting to determine whether ECM1 has other effects on
endothelial cells in addition to proliferation.
The association of ECM1 with tumor cells has a number of implications. First, the biological activity of ECM1 suggests that endothelial cells may have ECM1 receptors on their surface. In preliminary studies, we found that treatment of endothelial cells with ECM1a results in increased tyrosine phosphorylation, indicating the possible existence of a receptor for ECM1. Clearly, the identification of such a receptor would help in elucidating the signal transduction pathway. Second, ECM1 may represent an easily accessible marker for the presence of tumors. Because ECM1 is a secretory protein, the concentrations of this protein in the serum of patients may be correlated with the presence of certain types of tumors. Such a marker could prove to be very useful for assessing tumor load and may be of prognostic value in determining tumor recurrence after surgery.
In conclusion, the results of this study indicate that ECM1 has angiogenic properties and is produced by tumor cells, suggesting that it plays a role in tumor progression by stimulating tumor vascularization. If this is the case, it may be possible to slow or even reverse tumor progression by blocking the activity of ECM1 with competing factors or inhibitors.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
2 Jozef Merregaert and Lurong Zhang share senior
authorship.
Received for publication October 28, 1999.
Revision received September 27, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  E. Mikaelsson, A. H. Danesh-Manesh, A. Luppert, M. Jeddi-Tehrani, M.-R. Rezvany, R. A. Sharifian, R. Safaie, A. Roohi, A. Osterborg, F. Shokri, et al. Fibromodulin, an extracellular matrix protein: characterization of its unique gene and protein expression in B-cell chronic lymphocytic leukemia and mantle cell lymphoma Blood, June 15, 2005; 105(12): 4828 - 4835. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V Berno, D Porrini, F Castiglioni, M Campiglio, P Casalini, S M Pupa, A Balsari, S Menard, and E Tagliabue The 67 kDa laminin receptor increases tumor aggressiveness by remodeling laminin-1 Endocr. Relat. Cancer, June 1, 2005; 12(2): 393 - 406. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Mongiat, J. Fu, R. Oldershaw, R. Greenhalgh, A. M. Gown, and R. V. Iozzo Perlecan Protein Core Interacts with Extracellular Matrix Protein 1 (ECM1), a Glycoprotein Involved in Bone Formation and Angiogenesis J. Biol. Chem., May 2, 2003; 278(19): 17491 - 17499. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Hamada, W. H. I. McLean, M. Ramsay, G. H. S. Ashton, A. Nanda, T. Jenkins, I. Edelstein, A. P. South, O. Bleck, V. Wessagowit, et al. Lipoid proteinosis maps to 1q21 and is caused by mutations in the extracellular matrix protein 1 gene (ECM1) Hum. Mol. Genet., April 1, 2002; 11(7): 833 - 840. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
