| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1
* Zentrum Anatomie, Abteilung Histologie, Universität, Göttingen, D-37075 Göttingen, Germany; and
Max-Planck-Institut für Biochemie, D-82152 Martinsried, Germany
1Correspondence: Max-Planck-Institut für Biochemie, Am Klopferspitz 18A, D-82152 Martinsried, Germany. E-mail: TIMPL{at}biochem.mpg.de
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: binding assays • extracellular matrix • immunogold staining
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2
3)
. Perturbations of this balanced control seem to be
associated with several pathological conditions that are not yet well
defined at the molecular level. It would seem, however, that prevention
of endogenous inhibition of angiogenesis is essential for tumor growth
(4
, 5)
. A few angiogenesis inhibitors have in fact been
shown to have a strong therapeutic potential in experimental tumor
models (6
, 7)
. These endogenous inhibitors are a diverse
group of proteins, including the chemokine platelet factor 4
(8)
, prolactin (9)
, and the angiostatin part
of plasminogen (10)
, which require proteolytic activation.
Other putative inhibitors are typical extracellular matrix proteins and
include fibronectin (11)
, thrombospondin-1 (12
, 13)
, SPARC/BM-40/osteonectin (14)
, and the collagen
XVIII-derived endostatin (15)
. Again, proteolysis may be
required to generate an active inhibitory structure or for release in a
soluble endocrine form. Yet the diverse structures of these inhibitors
indicate that they are involved in different interactions, and their
release and role in homeostatic control are still not well defined.
The endostatin inhibitor (180 residues) is part of the larger
carboxyl-terminal globular domain NC1 (315 residues) of collagen XVIII
(16
, 17)
and was originally isolated as an inhibitor of
endothelial cell proliferation from hemangioendothelioma medium
(15)
. Recombinant endostatin from bacteria was largely
insoluble, but shown to efficiently arrest tumor growth in
vivo (7
, 15)
. A highly soluble recombinant mouse
endostatin and the NC1 domain were obtained from mammalian cells;
endostatin was shown by X-ray crystallography to be folded into a
compact globular structure with some relationship to C-type lectins
(18)
. Antibodies raised against endostatin demonstrated
its presence in serum, urine, and a large variety of tissues
(19)
. The tissue forms were mainly related in size to NC1,
whereas the circulating form corresponded more closely to endostatin.
Immunohistology of embryonic tissues demonstrated a strong reaction for
endostatin in many, but not all, blood vessels and some other basement
membrane zones. Further data showed similar in vitro binding
properties of NC1 and endostatin for the microfibrillar fibulins, but
higher affinities of NC1 for other extracellular ligands. This led us
to speculate that after the initial proteolytic release of the
procollagen-like NC1 domain, additional protease processing in a hinge
region is required to produce an endocrine form of endostatin similar
to that found in the circulation (19)
.
To examine further the features of endostatin related to homeostasis, we have now used immunoelectron microscopy and found a strong association with elastic fibers of arteries. In aorta and skin, the size corresponded to that of circulating endostatin. There was also a distinct colocalization of endostatin with fibulin-1, fibulin-2, and nidogen-2, which correlated with in vitro binding of endostatin to these proteins. This indicates that large vessels represent a rich reservoir of endostatin whose role remains to be identified.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, mouse
fibulin-1C (20)
and fibulin-2 (21)
, and human
nidogen-2 (22)
were obtained as recombinant products from
transfected mammalian cells. Recombinant human tropoelastin (rTE
26)
from bacteria and a corresponding rabbit antiserum (23
, 24)
were kindly supplied by J. Rosenbloom. Perlecan was purified
from a mouse tumor (25)
. Affinity-purified rabbit
antibodies against endostatin (19)
, fibulin-1
(20)
, fibulin-2 (21)
, nidogen-2
(22)
, and perlecan fragment III-3 (26)
were
described previously. A rat monoclonal antibody against mouse
endostatin was obtained by a lymph node injection protocol
(27)
in collaboration with Y. Ninomiya and colleagues
(unpublished results).
Protein ligand and immunological binding assays
Solid-phase binding assays to plastic-immobilized recombinant
mouse endostatin and affinity chromatography on a 1 ml heparin column
have been described (19)
. Immunoblotting of reduced tissue
extracts was performed with affinity-purified antibodies
(28)
. A radioimmune inhibition assay specific for mouse
endostatin was calibrated with recombinant endostatin for quantitative
measurements (19)
.
Tissue extractions and analysis
Freshly isolated aorta and skin from adult mice (0.22–0.26 g)
were homogenized in 1.5 ml 0.05 M Tris-HCl, pH 7.4, 0.1 M NaCl
(TBS)2
and split into three aliquots. One 0.5 ml aliquot (procedure A) was
extracted for 16 h at 4°C after addition of 10 mM EDTA and 1 mM
protease inhibitor Pefabloc (Boehringer Mannheim, Mannheim, Germany).
After centrifugation to obtain the EDTA buffer extracts, the residue
was extracted with 0.5 ml of the same buffer containing in addition 1%
Nonidet P-40, 0.5% deoxycholate, and 0.1% sodium dodecyl sulfate
(28)
for 4 h at 4°C (detergent extract). Another
0.5 ml aliquot of the TBS homogenate (procedure B) was supplemented
with 0.25 mg pancreatic elastase (Serva, Heidelberg, Germany) and
incubated for 16 h at 37°C in order to obtain an elastase
digest. The residue was then incubated (4 h, 37°C) in 0.5 ml TBS
containing 0.25 mg bacterial collagenase (Worthington, Freehold, N.J.).
The reverse order of collagenase (16 h) and elastase (4 h) was also
used to obtain further extracts from an additional 0.5 ml aliquot
(procedure C). In control experiments, recombinant endostatin was
treated separately with elastase or collagenase (16 h, 37°C) at an
enzyme substrate ratio of 1:30, which was approximately the same as the
ratio of protease to total protein in the tissue digestions.
Aliquots of the tissue extracts were hydrolyzed with 6 M HCl (16 h, 110°C) and analyzed on a Biotronic LC 3000 amino acid analyzer to determine the total protein content. The collagen content was estimated from the 4-hydroxyproline values, assuming a content of 10 mol %, as is typical for the fibrillar collagens, which represent the majority of the collagens present in these tissues.
Immunohistological analyses
Affinity-purified rabbit antibodies against endostatin (10
µg/ml) were used on frozen tissue sections of adult mice and their
binding was detected by fluorescence (Cy3)-conjugated goat anti-rabbit
immunoglobulin G (IgG) following previous protocols (28)
.
Negative controls were performed with normal rabbit IgG (10 µg/ml)
and showed no substantial staining (not shown).
Tissue pieces (~1 mm2) of aorta, large vessels,
kidney, liver, and skin of adult NMRI mice were fixed in 4%
paraformaldehyde and 0.5% glutaraldehyde for 15 min, dehydrated,
and embedded in the acrylic resin LR-Gold (London Resin Company,
London, U.K.). Ultrathin sections were obtained with a Reichert
ultramicrotome and collected on formvar coated nickel grids for
immunoelectron microscopy as described previously (29)
.
Affinity-purified goat anti-rat IgG (0.7 mg/ml) and goat anti-rabbit
IgG (1.0 mg/ml) were obtained from Medac (Hamburg, Germany) and used to
coat either 8 or 16 nm colloidal gold particles, following standard
protocols (30
, 31)
. Tissue sections were incubated at room
temperature (15 min) with phosphate-buffered saline pH 7.4 (PBS),
followed for 20 min with affinity-purified rabbit antibodies against
endostatin (6 µg/ml), fibulin-1 (6 µg/ml), fibulin-2 (17 µg/ml),
nidogen-2 (4 µg/ml), or perlecan (16 µg/ml) diluted in PBS.
Sections were rinsed in PBS and incubated (20 min) with 16 nm
gold-coupled anti-rabbit IgG (1:200 in PBS). After rinsing with water
and staining with uranyl acetate (15 min) and lead citrate (5 min),
sections were examined with a Zeiss EM 109 microscope.
For double labeling, a hybridoma medium containing rat monoclonal antibody against endostatin with an enzyme-linked immunoassay titer of 1:20,000 was used. The protocol was the same except that after treatment with the first and second antibody (see above), incubation was continued for 1 h with the monoclonal antibody (diluted 1:300, ~3 µg/ml), followed by 8 nm gold-coupled anti-rat IgG (20 min). Nonspecific binding of the gold probes to anionic sites in tissues was examined by incubation with the colloidal gold alone prior to or after coupling to anti-IgG. All controls proved to be negative (not shown).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. The data also indicated its
presence in adult tissues, which were now analyzed more comprehensively
by immunofluorescence (examples shown in Fig. 1
) in order to identify appropriate targets for immunoelectron
microscopy. Some of the strongest staining was observed in the aortic
media (Fig. 1A
) and included four or five lamellar
structures reminiscent of the multiple elastic sheets. The adventitia
showed only little staining, mainly in small vessels. A distinct
staining of large vessel walls was a general observation and included
portal veins and arteries as well as sinusoidal regions in the liver
(Fig. 1B
), large arteries and veins in kidney, heart, and
skeletal muscle, and, to a lesser extent, structures representing small
vessels. Endostatin staining was also observed in the renal mesangium,
Bowman's capsule, and along tubular basement membranes (Fig. 1C
). Staining of skin showed endostatin along the
dermal–epidermal junction, around hair follicles and small
vessels, and some additional extracellular deposits in the dermis (Fig. 1D
).
|
Immunogold staining shows endostatin association with elastic
fibers of large vessels
Based on the immunofluorescence data, a greater number of blood
vessels were examined in order to identify the ultrastructural
localization of endostatin (Fig. 2
). In aorta, there was a strong staining of the elastic fiber underneath
the endothelial cell layer (Fig. 2A
) and across all other
elastic sheets within the media (Fig. 2B
). Extracellular
labeling was found on top of amorphous elastic plaques as well as in
close proximity to microfibrillar deposits. No or only little staining
was observed in basement membranes adjacent to endothelial and smooth
muscle cells and within the cells. A distinct but less intense
endostatin staining was also found in elastic fibers of several large
muscular arteries (Fig. 2C
) and in the extracellular matrix
containing sparse elastic fibers of veins (Fig. 2D
) and
venules of variable size. A more quantitative analysis of identical
areas of these vessels demonstrated 20–25 gold particles on the
elastic sheets and 3–5 on intracellular regions when compared to 1–3
gold particles in the negative controls, indicating that the
differences are significant. Some but not all arterioles and
capillaries examined also showed a weak labeling for endostatin. It was
found in the vessel lumen. where it likely represents the circulating
form of endostatin (19)
, but was more prominent in
microfibrillar regions underneath the vessel walls (Fig. 2E
). Again, no staining was found for basement membranes.
|
As expected, the same elastic fibers as shown in Fig. 2
could be
strongly stained with antibodies to tropoelastin (data not shown). They
were also labeled with antibodies against several extracellular matrix
proteins known to be microfibrillar and/or vessel wall components
(20
21
22
, 32
, 33)
in double staining experiments, as shown
for the aortic media (Fig. 3
). Staining was observed for the elastic fiber components fibulin-2
(Fig. 3A
), fibulin-1 (Fig. 3B
), nidogen-2 (Fig. 3C
), and the heparan sulfate proteoglycan perlecan (Fig. 3D
), whereas versican and laminin-1 could not be detected.
Double staining with a monoclonal antibody to endostatin showed a
frequent colocalization within a distance of 10–20 nm with fibulin-2
and nidogen-2 (Fig. 3A, C
). Such colocalizations were less
obvious for fibulin-1 and negligible for perlecan (Fig. 3B, D
).
|
A few more tissue structures were studied by immunogold staining in
order to get an overview of further ultrastructural associations of
endostatin. In skin, there was a weak to moderate labeling of dermal
collagen fibers and of the papillary dermis underneath the
dermal–epidermal basement membrane (Fig. 2F
). The latter is
known to contain many microfibrils composed of fibronectin, fibulin-2,
and fibrillin (34)
. Staining could not be detected in the
dermal regions of sparse elastic fibers. A typical basement membrane
staining for endostatin was found in the proximal tubules of kidney
(Fig. 4
A), whereas glomerular basement membranes and the podocytes
were only weakly labeled. In the liver, besides the large vessels,
endostatin could be localized primarily to the rough endoplasmic
reticulum and Golgi complex of hepatocytes (Fig. 4B).
|
Solubilization and properties of endostatin from aorta and skin
The immunogold staining of aorta and skin suggested that they may
differ in the content and molecular associations of endostatin. Both
tissues were therefore subjected to three different
extraction/digestion protocols: solubilization by EDTA-containing
neutral buffer, followed by detergents (procedure A), and two
consecutive proteolytic digestions by either elastase, followed by
collagenase (procedure B), or collagenase, followed by elastase
(procedure C). All extracts were analyzed for their total protein and
collagen contents and for the amount of solubilized endostatin antigen
by radioimmunoassay (Table 1
). For aorta, the most efficient extraction of endostatin was obtained
using procedure A (~80 µg/g wet tissue), which corresponded to
0.5% of the total proteins solubilized. A comparable extraction of
other mouse tissues (brain, kidney, liver, testis) by EDTA-containing
buffer alone yielded soluble endostatin in the range 2–10 µg/g wet
tissue when compared to 38 µg/g for aorta. The two proteolytic
protocols solubilized much more total protein and collagen from aorta,
but only 15–30% of the endostatin when compared to procedure A.
|
The skin contained less extractable endostatin (13 µg/g wet tissue)
than aorta when analyzed by procedure A, and again yielded lower
amounts of endostatin by the proteolytic procedures B and C. A
comparable extensive treatment of recombinant endostatin by either
collagenase or elastase did not cause any change in electrophoretic
mobility or any loss of antigenicity in the radioimmune inhibition
assay. This confirmed previous observations of a high stability against
neutral proteases (19)
and demonstrated that the lower
yields of endostatin in the proteolytic extracts are not due to
degradation.
The nature of the endostatin antigen in the various extracts were
analyzed by immunoblotting after electrophoretic separation (Fig. 5
). Most of the aorta extracts contained distinct components of 25 and 22
kDa, the latter being the size of recombinant endostatin. No larger
material with the size of carboxyl-terminal domain NC1 (35–38 kDa)
could be detected, although this has previously been shown to be
prominent in muscle, testis, and liver extracts (19)
. The
22 kDa form dominated in the skin extracts, some of which also
contained a 38 and 60 kDa band that could represent NC1 and a larger
partially degraded collagen XVIII fragment. A mixture of the EDTA and
detergent extracts from aorta was also subjected to analytical heparin
affinity chromatography, followed by radioimmunoassay quantitation of
endostatin. This demonstrated efficient binding (>90%) and elution
from the column at 0.34 M NaCl, similar to recombinant endostatin
(19)
, and indicated that the solubilized aorta endostatin
is not complexed to a ligand that would interfere with heparin binding.
|
Binding of elastic fiber proteins to recombinant endostatin
Previous analysis by surface plasmon resonance assay demonstrated
a distinct binding of endostatin to fibulins (19)
. As a
result of the colocalization data presented here, we examined these and
other ligands in solid-phase assays (Fig. 6
). This demonstrated a relatively strong binding of soluble fibulin-2,
nidogen-2, and fibulin-1 to immobilized endostatin (half-maximal at
1–10 nM) and a 10- to 100-fold weaker reaction with nidogen-1 and
perlecan. A similar analysis with soluble tropoelastin showed a strong
binding to albumin-coated control plastic wells (half-maximal at 2 nM)
and only a slight increase in the presence of immobilized endostatin.
No endostatin binding to immobilized tropoelastin was observed in
solid-phase and surface plasmon resonance assays, indicating a lack of
any substantial affinity.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
36
37)
, but it is unclear whether the two domains
colocalize at the ultrastructural level. We used immunogold staining to
clearly identify endostatin as an extracellular matrix component as
well as finding it within some intracellular compartments.
An unexpected finding was the strong association of endostatin with all
elastic sheets in aorta. A distinct but less intense staining was
observed in the elastic fibers of large arteries and veins. There was
also a frequent colocalization with microfibrillar structures in the
elastic lamellae of the aorta. Such structures also exist without
elastin in the papillary dermis (34)
and other
extracellular locations. Intracellular endostatin was particularly
prominent in hepatocytes, where it seems to be associated with the
rough endoplasmic reticulum and Golgi complex and could thus contribute
the plasma form of endostatin, as previously suggested
(19)
. This observation correlated well with a high
expression of collagen XVIII in normal and fibrotic human liver
(36)
. Endostatin can, however, also be a basement membrane
component, as shown for the proximal tubular epithelium of kidney.
Thinner basement membranes in the close proximity to endothelial and
smooth muscle cells of vessel walls could not be labeled for
endostatin, but were negative for ubiquitous basement membrane proteins
such as perlecan and nidogen-1 (data not shown). Therefore we cannot
exclude that the fixation and surface immunogold staining protocol used
here is not optimal for these basement membranes. Small capillaries
that contained endostatin in the vessel lumen and more remote
microfibrils also lacked vessel wall staining.
Recombinant endostatin produced and secreted by mammalian cells is a
highly soluble and crystallizable protein (18
, 19)
. The
localization of its tissue analogs to three different extracellular
ultrastructures (amorphous elastic plaques, microfibrils, basement
membranes) strongly suggests that endostatin is kept in place there by
interactions with other structural extracellular ligands. Solid-phase
binding assays identified fibulin-2, nidogen-2, and fibulin-1 as three
potential ligands, whereas other interactions with nidogen-1 or
perlecan may be too weak to be relevant. These observations correlated
well with a distinct colocalization of endostatin with fibulin-2 and
nidogen-2 in the elastic sheets of the aortic media. The double
immunogold labels were often close enough (10 nm) to be consistent with
molecular contacts. These postulated contacts will require additional
biochemical characterization after chemical cross-linking of such
complexes.
Elastic sheets and fibers are prominent in aorta and other large
arteries but are also found in lung, skin, elastic cartilage, and some
special tissues. They consist of a highly insoluble, covalently
cross-linked network of elastin monomers (tropoelastin) that is
frequently associated with fibrillin-containing microfibrils and some
associated proteins (38
, 39)
. More recent additions to
this complexity include fibulin-1 (40)
and fibulin-2
(41)
in dermal elastic fibers and, as shown here, in the
elastic sheets of aorta. Fibulin-2 was also colocalized to microfibrils
composed of either fibronectin or fibrillin (28
, 34
, 41)
.
Besides endostatin, nidogen-2 and perlecan were also identified as
novel components in the elastic structures of aorta by immunogold
staining. Nidogen-2 is a recently discovered extracellular protein that
has a length of ~50 nm, binds to collagen IV and perlecan, and is
structurally and functionally related to the ubiquitous basement
membrane protein nidogen-1 (22)
. Nothing is known about
its supramolecular organization and ultrastructural tissue
localization. Recent data, however, indicate a strong affinity for
tropoelastin (T. Sasaki, W. Göhring, N. Miosge, J.
Rosenbloom, R. Timpl, unpublished results), which could explain the
nidogen-2 localization in aorta. The heparan sulfate proteoglycan
perlecan is a well-characterized component of basement membranes and
other interstitial spaces with many potential functions (33
, 42)
. It has not yet been identified in elastic structures, but
our data agree with previous observations that they contain
proteoglycans (39)
. Perlecan may not be a relevant ligand
for endostatin in the elastic parts of aorta because of its presumably
low content, but it could serve this purpose in basement membranes.
Extraction and initial biochemical characterization of endostatin from
various tissues (19)
identified aorta as the richest
source of the potential endothelial cell inhibitor known so far.
Furthermore, the size of endostatin in aorta and skin extracts was in
the range 22–25 kDa, which is similar to that of endostatin in human
plasma but distinctly smaller than the major variants (35–38 kDa)
found in several other tissues (19)
. Full-length collagen
XVIII was identified in tissues and cell cultures by antibodies against
the amino-terminal domain NC11 as a 180–200 kDa band (35
, 37)
, which we could not detect in aorta and skin by the
endostatin antibodies. This indicates a complete proteolytic release of
the carboxyl-terminal endostatin domain in both tissues, but we cannot
entirely exclude the possibility that a heparan sulfate-substituted
form of collagen XVIII (37
, 43)
may have escaped our
detection. Based on previous data (19)
, we postulated that
processing of collagen XVIII starts with the release of the
carboxyl-terminal NC1 domain (38 kDa), as known for other procollagens,
followed by a second proteolytic cleavage in a hinge region, which then
releases endostatin with some variability in size (22–25 kDa). This
process seems to be almost complete in aorta and skin and may not even
require the initial proteolytic step.
In the tissue extraction of endostatin, a combination of neutral
buffers with a chelating agent and detergents yielded the highest
amounts from aorta and skin. This could reflect the small size of the
endostatin antigens present there, but also implicates the involvement
of divalent cations and hydrophobic interactions in the binding of
endostatin to the extracellular matrix. In this context, it is of
interest that human endostatin contains a single zinc atom
(44)
, as also found for mouse endostatin (T. Sasaki, E.
Hohenester, R. Timpl, unpublished results). Whether the removal of zinc
or of calcium from the potential fibulin and nidogen-2 ligands
(20
, 22
, 28)
is responsible for solubilization by EDTA
remains to be analyzed. A more extensive tissue solubilization by a
combination of elastase and collagenase treatments was less efficient
for the extraction of endostatin, which was stable against both
proteases. This agreed with our observations that endostatin and
tropoelastin have no remarkable affinity for each other. The data also
support our interpretation that most of the endostatin is already
released from collagen XVIII, since no 38 kDa NC1 components could be
detected by immunoblotting after collagenase treatment.
Relatively little is known about the biological role of endogenous
endostatin except for indications that it could inhibit endothelial
cell growth and interfere with tumor growth in experimental mouse
models (7
, 15)
. Yet its abundance in many extracellular
locations strongly suggests a role. We interpret our observation of a
strong association with elastic structures of aorta as being mainly
mediated by associated microfibrillar components containing fibulin-2
and/or nidogen-2. Other possible associations still remain to be
identified. One major unresolved questions is whether domain NC1 itself
has inhibitory activity or whether it requires further conversion to
endostatin, which also decreases the potential for protein interactions
(19)
. Since the endostatin in aorta is near to the
endothelial cell layer, one may speculate that this reservoir prevents
deleterious angiogenesis from the luminal side that may be initiated by
pathological triggers to the cells. The analysis of such possibilities
and other parameters will become instrumental in understanding the
endogenous control of angiogenesis inhibition.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication February 15, 1999. Revised for publication April 23, 1999.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
1 chain of mouse type XVIII collagen, partial structure of the corresponding gene, and comparison of the 1(XVIII) chain with its homologue, the 1(XV) collagen chain. J. Biol. Chem. 269,13929-13935 chains of human type IV collagen. Histochem. Cell Biol. 104,267-275[Medline]
1-chain and the E1 and E8 domains of laminin-1 in mouse kidney using monoclonal antibodies to establish the orientation of laminin-1 within basement membranes. J. Histochem. Cytochem. 43,675-680[Abstract]
This article has been cited by other articles:
|
|
 |
|
  B. Doyle and N. Caplice Plaque Neovascularization and Antiangiogenic Therapy for Atherosclerosis J. Am. Coll. Cardiol., May 29, 2007; 49(21): 2073 - 2080. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C.-H. Yi, D. J. Smith, W. W. West, and M. A. Hollingsworth Loss of Fibulin-2 Expression Is Associated with Breast Cancer Progression Am. J. Pathol., May 1, 2007; 170(5): 1535 - 1545. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Piwnica, I. Fernandez, N. Binart, P. Touraine, P. A. Kelly, and V. Goffin A New Mechanism for Prolactin Processing into 16K PRL by Secreted Cathepsin D Mol. Endocrinol., December 1, 2006; 20(12): 3263 - 3278. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Schaffhauser, T. Veikkola, K. Strittmatter, H. Antoniadis, K. Alitalo, and G. Christofori Moderate antiangiogenic activity by local, transgenic expression of endostatin in Rip1Tag2 transgenic mice J. Leukoc. Biol., October 1, 2006; 80(4): 669 - 676. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Wenzel, A. Schmidt, K. Reimann, J. Hescheler, G. Pfitzer, W. Bloch, and B.K. Fleischmann Endostatin, the Proteolytic Fragment of Collagen XVIII, Induces Vasorelaxation Circ. Res., May 12, 2006; 98(9): 1203 - 1211. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. V. Lee, J. C. Rodriguez-Manzaneque, S. N.-M. Thai, W. O. Twal, A. Luque, K. M. Lyons, W. S. Argraves, and M. L. Iruela-Arispe Fibulin-1 Acts as a Cofactor for the Matrix Metalloprotease ADAMTS-1 J. Biol. Chem., October 14, 2005; 280(41): 34796 - 34804. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 X. Zeng, J. Chen, Y. I. Miller, K. Javaherian, and K. S. Moulton Endostatin binds biglycan and LDL and interferes with LDL retention to the subendothelial matrix during atherosclerosis J. Lipid Res., September 1, 2005; 46(9): 1849 - 1859. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Gersdorff, E. Kohfeldt, T. Sasaki, R. Timpl, and N. Miosge Laminin {gamma}3 Chain Binds to Nidogen and Is Located in Murine Basement Membranes J. Biol. Chem., June 10, 2005; 280(23): 22146 - 22153. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Y. Zhang, E. G. Teggatz, A.-P. Zou, W. B. Campbell, and P.-L. Li Endostatin uncouples NO and Ca2+ response to bradykinin through enhanced O2-{middle dot} production in the intact coronary endothelium Am J Physiol Heart Circ Physiol, February 1, 2005; 288(2): H686 - H694. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A.-K. Olsson, I. Johansson, H. Akerud, B. Einarsson, R. Christofferson, T. Sasaki, R. Timpl, and L. Claesson-Welsh The Minimal Active Domain of Endostatin Is a Heparin-Binding Motif that Mediates Inhibition of Tumor Vascularization Cancer Res., December 15, 2004; 64(24): 9012 - 9017. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. S. Moulton, B. R. Olsen, S. Sonn, N. Fukai, D. Zurakowski, and X. Zeng Loss of Collagen XVIII Enhances Neovascularization and Vascular Permeability in Atherosclerosis Circulation, September 7, 2004; 110(10): 1330 - 1336. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Yu, K. S. Moulton, M. K. Khan, S. Vineberg, E. Boye, V. M. Davis, P. E. O'Donnell, J. Bischoff, and D. S. Milstone E-selectin is required for the antiangiogenic activity of endostatin PNAS, May 25, 2004; 101(21): 8005 - 8010. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. E. Daly, A. Makris, M. Reed, and C. E. Lewis Hemostatic Regulators of Tumor Angiogenesis: A Source of Antiangiogenic Agents for Cancer Treatment? J Natl Cancer Inst, November 19, 2003; 95(22): 1660 - 1673. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Yi, T. Sakai, R. Fassler, and E. Ruoslahti Antiangiogenic proteins require plasma fibronectin or vitronectin for in vivo activity PNAS, September 30, 2003; 100(20): 11435 - 11438. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. H. DEININGER, W. A. WYBRANIETZ, F. T.C. GRAEPLER, U. M. LAUER, R. MEYERMANN, and H. J. SCHLUESENER Endothelial endostatin release is induced by general cell stress and modulated by the nitric oxide/cGMP pathway FASEB J, July 1, 2003; 17(10): 1267 - 1276. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Miosge, T. Simniok, P. Sprysch, and R. Herken The Collagen Type XVIII Endostatin Domain Is Co-localized with Perlecan in Basement Membranes In Vivo J. Histochem. Cytochem., March 1, 2003; 51(3): 285 - 296. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Kato, J.-H. Chang, and D. T. Aza |
), nidogen-2
(
), fibulin-1C (•), nidogen-1 (
), and perlecan (
).