| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Cell Biology, Harvard Medical School, Boston, Massachusetts, USA
1 Correspondence: Department of Cell Biology, Harvard Medical School, 240 Longwood Ave., Boston, 02115, MA, USA. E-mail: Alexander_Marneros{at}hms.harvard.edu and Bjorn_Olsen{at}hms.harvard.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONCOLLAGEN XVIII/ENDOSTATIN GENE...ORGANIZATION OF COLLAGEN XVIII...ABNORMAL VASCULATURE DUE TO...AGE-DEPENDENT LOSS OF VISION...EVIDENCE FOR A FUNCTIONALLY...ABNORMALITIES OF THE IRIS...MIGRATION OF PIGMENTED IRIS...HAS COLLAGEN XVIII/ENDOSTATIN...SUMMARYREFERENCES |
|---|
Key Words: basement membrane • retinal pigment epithelium • age-related macular degeneration • hydrocephalus • heparan sulfate proteoglycan
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONCOLLAGEN XVIII/ENDOSTATIN GENE...ORGANIZATION OF COLLAGEN XVIII...ABNORMAL VASCULATURE DUE TO...AGE-DEPENDENT LOSS OF VISION...EVIDENCE FOR A FUNCTIONALLY...ABNORMALITIES OF THE IRIS...MIGRATION OF PIGMENTED IRIS...HAS COLLAGEN XVIII/ENDOSTATIN...SUMMARYREFERENCES |
|---|
). Collagen XVIII/endostatin is a recently identified component of almost all epithelial and endothelial BMs (2
3
4
5
6)
. This collagen is a heparan sulfate proteoglycan (7)
and contains 10 collagenous (COL) domains that are interrupted and flanked by noncollagenous domains (NC) (5
, 8)
. A proteolytic fragment of the C-terminal noncollagenous domain (NC1), termed endostatin, has been shown to have anti-angiogenic activity in vitro and in vivo (9
, 10)
. However, lack of collagen XVIII and endostatin in mice and humans does not increase angiogenesis in major organs, suggesting that this collagen or its proteolytic derivative endostatin is not a critical negative regulator of angiogenesis during development and postnatal growth (3
, 11)
.
Instead, collagen XVIII/endostatin may be just one component of several that are part of the balance between pro-angiogenic and anti-angiogenic regulators. Lack of such a single component involved in the dynamic angiogenic remodeling processes may reveal no obvious alteration in angiogenesis unless it is a critical regulator. For example, besides endostatin as a fragment of collagen XVIII, fragments of several other BM proteins have been implicated to affect angiogenesis as well. Endorepellin, derived from the C-terminus of perlecan, has been shown to inhibit angiogenesis (12)
. This fragment causes a disassembly of the actin cytoskeleton and focal adhesions in endothelial cells (13)
. Other examples are different fragments of type IV collagen that have been reported to inhibit angiogenesis in vitro and in vivo (reviewed in ref. 10
). Factors regulating angiogenesis may have some tissue specificity as well. For example, PEDF (pigment epithelium-derived factor) is a major anti-angiogenic factor in the eye that can prevent the growth of vessels into the avascular ocular compartments, such as cornea or vitreous, and can also inhibit experimental choroidal neovascularization (14
, 15)
. An anti-angiogenic role for PEDF has been reported in some other tissues such as pancreas or prostate as well (16)
.
Evidence that collagen XVIII/endostatin is indeed a regulator of angiogenesis, but not a critical one, has recently been provided through aortic explant culture experiments using Col18a1–/– and wild-type aortas in serum-free medium conditions (lacking important factors found in serum that are involved in angiogenicremodeling processes). In these experiments, lack of collagen XVIII/endostatin resulted in increased angiogenesis, which could be reversed by addition of endostatin to these cultures (17)
. In contrast, a difference in the angiogenic response between wild-type and Col18a1–/– mice was not detected in other in vitro models of angiogenesis (like the metatarsal vascular outgrowth assay) that require the addition of serum.
Use of anti-angiogenic proteins or peptides in anti-tumor treatments may achieve greater pharmacologic potency when several anti-angiogenic factors are combined and not given individually. Viewing collagen XVIII/endostatin as one of several regulators of angiogenesis may help explain tissue-specific phenotypes in humans and mice lacking this collagen despite its expression in almost all BMs. Recent evidence has demonstrated a tissue-specific requirement of this collagen, particularly in the eye and brain (3
, 18
19
20
21)
. It is possible that other factors may compensate for the loss of collagen XVIII/endostatin in most tissues while such compensation is lacking in the eye or brain, resulting in a tissue-specific phenotype. Several examples of tissue-specific phenotypes due to a mutant gene that is expressed in a wide variety of tissues has been reported in the literature, such as the ocular disease Sorsby’s fundus dystrophy caused by mutations in the TIMP3 gene (22)
.
Inactivating mutations in the gene for collagen XVIII result in a progressive attenuation of vision in mice and humans, with distinct developmental and age-dependent abnormalities in ocular structures. Characterization of ocular defects in mice lacking collagen XVIII/endostatin revealed striking histopathological similarities to common human ocular diseases, such as early age-related macular degeneration (ARMD) or human pigment dispersion syndrome. The findings emphasize the importance of this collagen for the proper function of ocular BMs and for maintenance of ocular epithelial metabolism and visual sensitivity. Furthermore, a predisposition for hydrocephalus formation has been reported in a Col18a1 null mouse strain (21)
, providing evidence for a more general defect in BMs that is not restricted to ocular BMs. Indeed, an altered BM ultrastructure has been noted in several different tissues in this Col18a1 null mouse strain (21)
.
|
COLLAGEN XVIII/ENDOSTATIN GENE MUTATIONS IN KNOBLOCH SYNDROME |
|---|
|
TOPABSTRACTINTRODUCTIONCOLLAGEN XVIII/ENDOSTATIN GENE...ORGANIZATION OF COLLAGEN XVIII...ABNORMAL VASCULATURE DUE TO...AGE-DEPENDENT LOSS OF VISION...EVIDENCE FOR A FUNCTIONALLY...ABNORMALITIES OF THE IRIS...MIGRATION OF PIGMENTED IRIS...HAS COLLAGEN XVIII/ENDOSTATIN...SUMMARYREFERENCES |
|---|
.
Knobloch syndrome is an autosomal recessive disorder characterized by the occurrence of vitreoretinal degeneration with retinal detachment, high myopia, macular degeneration, and occipital encephalocele (24)
. Ocular abnormalities display clinical variability and may include congenital cataracts, iris abnormalities, or lens subluxation in some patients. Besides the characteristic occipital encephalocele, further extraocular findings in Knobloch syndrome patients are rare and not typical of this syndrome. However, the eye findings are severe and regularly lead to blindness at young age (11)
.
Family members of the consanguineous Brazilian family with Knobloch syndrome have a homozygous mutation at the AG consensus sequence at the 3' end of intron 1 in COL18A1, whereas obligate carriers of the disease allele are heterozygous for this mutation. The mutation leads to skipping of exon 2 and the creation of a premature termination codon within exon 4 of the COL18A1 transcript (11)
.
The COL18A1 gene has two promoters: one upstream of exon 1, the second upstream of exon 3; transcription from these promoters results in distinct isoforms of collagen XVIII. The promoter upstream of exon 1 yields a transcript that contains exons 1, 2 and exons 4-43, but excludes exon 3, encoding a short isoform of the protein with a 1336 residue chain (NC11-303). The second promoter upstream of exon 3 yields two longer isoforms: NC11-493 (encoded by a transcript using a splice site within exon 3) and NC11-728 (including an additional 235 residues encoded by the sequence between the splice site within exon 3 and the splice site at the 3' end of exon 3) (25
, 26)
(Fig. 1
). The three isoforms of collagen XVIII differ in their N-terminal noncollagenous domain (NC11), with the longest human isoform (NC11-728) containing a cysteine-rich frizzled-like motif of unknown function. Expression analysis in human retinal tissue by RT-PCR identified the short isoform (NC11-303) and the longest isoform (NC11-728) of collagen XVIII to be expressed in this tissue.
|
The exon 2 skipping mutation in the consanguineous Brazilian family with Knobloch syndrome affects only the short isoform of collagen XVIII, whereas the long isofoms are not altered. Notably, affected members in this family lost their vision after 
20 years of age. In contrast, Knobloch syndrome patients with mutations that are predicted to cause the deficiency of all isoforms of collagen XVIII lost their vision in early childhood already (25)
. In Knobloch syndrome patients premature stop codons have been identified in exons 13, 24, 38, 40, and 42 of the COL18A1 gene (25
, 27)
. These patients do not differ clinically from Knobloch syndrome patients with mutations that affect only the short collagen XVIII isoform, with the exception of more severe ocular defects and earlier onset of vision loss mainly due to retinal detachment. These observations imply that the short (NC11-303) and the longest isoform (NC11-728), both being expressed in the eye, are important for maintenance of ocular structures and for vision.
Collagen XVIII isoforms can be detected in a wide variety of tissues. The short isoform is expressed in almost all vascular and epithelial BMs, whereas the long isoforms are highly expressed in the liver (2
, 26
, 28)
. It is therefore surprising that no major abnormalities in organs other than the eye (and the posterior part of the skull) are features of Knobloch syndrome in patients with mutations that affect all collagen XVIII isoforms. This suggests a particular requirement for collagen XVIII in eye tissues.
It is important to emphasize that considerable clinical overlap exists regarding ocular defects between Knobloch syndrome and other familial disorders (Stickler, Wagner, or Marshall syndromes) with mutations in collagens II and XI, components of the vitreous (26
, 29
30
31)
. Evidence for genetic heterogeneity in Knobloch syndrome with no mutations in the COL18A1 gene has been reported as well (25
, 27)
. Thus, mutations in different genes may result in very similar ocular abnormalities, probably due to common pathogenetic mechanisms.
No donor eye from a patient with Knobloch syndrome has been available so far, and thus histological ocular abnormalities resulting from the absence of collagen XVIII in humans remain unknown. Since inactivating mutations in the COL18A1 gene lead to Knobloch syndrome, it is likely that mice that are null for Col18a1 are a model for this syndrome and may provide insights into the pathogenetic mechanisms leading to the clinical observations in Knobloch syndrome patients (Table 1
).
|
|
ORGANIZATION OF COLLAGEN XVIII IN THE BM ZONE AND ABNORMALITIES OF BMs IN Col18a1 NULL MICE |
|---|
|
TOPABSTRACTINTRODUCTIONCOLLAGEN XVIII/ENDOSTATIN GENE...ORGANIZATION OF COLLAGEN XVIII...ABNORMAL VASCULATURE DUE TO...AGE-DEPENDENT LOSS OF VISION...EVIDENCE FOR A FUNCTIONALLY...ABNORMALITIES OF THE IRIS...MIGRATION OF PIGMENTED IRIS...HAS COLLAGEN XVIII/ENDOSTATIN...SUMMARYREFERENCES |
|---|
. This labeling is consistent with a localization of collagen XVIII in epithelial and endothelial BMs (2
, 26)
. Immuno-EM labeling with antibodies against the N-terminal NC11 domain and the C-terminal endostatin domain of collagen XVIII demonstrated a polarized orientation of this collagen within the BM zone, with the N-terminal domain being localized in the sub-lamina densa and the endostatin domain being part of BM molecular networks within the lamina densa (18)
(Fig. 2
). Colocalization of the C-terminal domain of collagen XVIII (using antibodies against the endostatin or NC1(XVIII) domain) with perlecan and laminin in the BM in immuno-EM double-labeling experiments is consistent with data from solid-phase binding assays showing protein interactions of endostatin with perlecan and laminin (32
, 33)
. A recent report suggests that an altered interaction of endostatin with laminin may contribute to the phenotype in a patient with Knobloch syndrome (11)
. Interaction of the endostatin domain with laminin has been shown in vitro to affect cell functions in many ways. For example, an oligomerization-dependent promigratory activity of the C-terminal NC1(XVIII)/endostatin domain of collagen XVIII (34)
is mediated in part through laminin in the BM (33)
. A monoclonal antibody that blocks the interactions between endostatin and laminin was able to inhibit the motogenic activity of endostatin oligomers (33)
. Sites of interaction between laminin and oligomeric endostatin include the N-terminal regions of all three laminin chains. These data collectively demonstrate that the C-terminal domain of collagen XVIII is part of BM molecular networks and that some of the effects observed for the endostatin domain involve interactions with laminin.
|
The endostatin domain may mediate biological effects through interactions with BM heparan sulfate proteoglycans such as perlecan or cell surface heparan sulfate proteoglycans such as glypicans. Direct binding of endostatin to perlecan has been demonstrated in binding assays (32)
, and close colocalization between endostatin and perlecan has been observed in BMs (18
, 35)
. Cell surface glypicans have been identified as low-affinity receptors for endostatin, with their heparan sulfate side chains being critical for endostatin binding (36)
. These data are supported by the crystal structure of endostatin, which revealed a basic heparin binding site formed by arginine residues (37
, 38)
. This heparin binding site has been shown to mediate endostatin’s anti-angiogenic effects in experiments where angiogenesis was induced with FGF-2 (39)
. Proper binding of endostatin to heparin and heparan sulfate requires the presence of divalent cations. The addition of ZnCl2 to endostatin enhanced its binding to heparan sulfate as well as its anti-proliferative effect on endothelial cells, suggesting that this anti-proliferative activity is mediated by the binding of endostatin to heparan sulfate (38)
.
The intracellular effects of endostatin binding to endothelial cells have recently been investigated in genome-wide expression profiling experiments (40)
. In this study, a large cluster of genes involved in endothelial cell functions were altered in their expression after adding endostatin to primary human endothelial cells, with an up-regulation of several known anti-angiogenic factors and down-regulation of proangiogenic factors. Net effects of endostatin on cultured endothelial cells include altered proliferation, cell migration, and apoptosis (10)
. The data support the hypothesis that endogenous endostatin is involved in the physiologic regulation of blood vessel formation.
The N-terminal noncollagenous domain of collagen XVIII (NC11) localizes to areas of the sub-lamina densa of BM zones (Fig. 2)
. This is the region where fibrillar structures are detected, suggesting a potential contribution of this collagen in anchoring the BM to underlying structures. The polarized orientation of collagen XVIII molecules was found in all ocular BMs and in extraocular BMs, with the exception of BMs lining the vitreous, such as the ILM of the retina or the posterior ciliary body BM. In these BMs, the N-terminal noncollagenous collagen XVIII domain (NC11) localized particularly to areas where vitreal fibrils inserted into the BM. Consistent with this localization, a reduced number of vitreal collagenous fibrils inserting into the ILM was observed in Col18a1–/– mice (3)
. This observation supports the hypothesis of a potential anchorage function of the N-terminal noncollagenous domain of collagen XVIII (NC11). This anchorage function may be mediated by the heparan sulfate glycosaminoglycan side chains of this domain (7)
.
Labeling for the N-terminal collagen XVIII domain was also detected at the insertion sites of zonular fibrils into the posterior ciliary body BM and the lens capsule. Some Knobloch syndrome patients with inactivating collagen XVIII mutations have lens subluxation (11
, 25)
, possibly due to a weakened anchorage of the zonular fibrils into the posterior ciliary body BM or the lens capsule. Since the zonule consists mainly of fibrillin-1, it is possible that the N-terminal domain of collagen XVIII interacts with fibrillin-1 or fibrillin-1-containing protein complexes. Lens subluxation, as observed in some Knobloch syndrome patients, is a clinical hallmark of Marfan syndrome with gene mutations affecting fibrillin-1 function (41)
.
|
ABNORMAL VASCULATURE DUE TO THE LACK OF COLLAGEN XVIII/ENDOSTATIN |
|---|
|
TOPABSTRACTINTRODUCTIONCOLLAGEN XVIII/ENDOSTATIN GENE...ORGANIZATION OF COLLAGEN XVIII...ABNORMAL VASCULATURE DUE TO...AGE-DEPENDENT LOSS OF VISION...EVIDENCE FOR A FUNCTIONALLY...ABNORMALITIES OF THE IRIS...MIGRATION OF PIGMENTED IRIS...HAS COLLAGEN XVIII/ENDOSTATIN...SUMMARYREFERENCES |
|---|
. Some clinical data have further implied an anti-tumor effect of endostatin based on a correlation of a reduced tumor incidence with increased endostatin serum concentrations in patients with trisomy 21 (who have 3 copies of the COL18A1 gene) (42)
. However, mice that lack collagen XVIII/endostatin are viable and fertile and do not display obvious vascular abnormalities or altered tumor growth characteristics (3)
. Thus, physiological levels of collagen XVIII/endostatin do not seem to be critical for the regulation of angiogenesis throughout development and postnatal growth.
However, careful examination of the ocular vasculature in Col18a1–/– mice did reveal developmental defects in hyaloid vessel regression and in the outgrowth and patterning of the retinal vasculature (3
, 43
, 44)
. Fluorescent angiography demonstrated abnormal retinal vessels in young Col18a1–/– mice (19)
. Retinal arteries were perfused despite irregular bending of major vessels, and no evidence for vascular leakage was found. The abnormal retinal vasculature in these mutant mice does not seem to affect retinal function and morphology, since no obvious histological retinal abnormalities were found in young mutant mice and electroretinograms (ERGs) were normal in these mice (18)
.
The abnormal retinal vascular pattern in Col18a1–/– mice may be a consequence of cell migration defects. During development of the retinal vasculature, endothelial cell migration is guided by VEGF-expressing astrocytes that form a preexisting astrocytic template. Astrocytes enter the eye from the optic nerve, proliferate, and migrate across the inner retina in advance of developing vessels to form a scaffold with guidance function for migrating endothelial cells (45)
. Such retinal vascular endothelial cell guidance is mediated by adhesion molecules, which are known to be involved in neuronal cell guidance as well. Since it has been shown that the C-terminal NC1 domain of collagen XVIII has a promigratory activity (34)
and that the lack of this domain leads to cell migration and axon guidance defects in Caenorhabditis elegans (46)
, it is tempting to speculate that lack of collagen XVIII may affect to some extent astrocytic or endothelial cell migration in the developing retina. Further support for a role of collagen XVIII/endostatin in cell migration comes from the observation of cerebral malformations in some Knobloch syndrome patients, possibly due to defects in neuronal cell migration (47)
.
Alternatively, abnormal retinal vessels may be a consequence of the delayed regression of hyaloid vessels in mutant mice. Hyaloid vessels along the ILM (vasa hyaloidea propria) had completely regressed in wild-type mice at day 16 after birth, but could still be seen in Col18a1–/– mice at that time, and were often enlarged in diameter (3)
. The hyaloid vessels of the eye normally regress according to a developmental program (48)
, mediated by macrophage-induced apoptosis (49
, 50)
. Hypoxia in the peripheral retina, possibly caused by the onset of neuronal activity and hyaloid vessel regression, induces up-regulation of VEGF expression in astrocytes, and thereby stimulates the outgrowth of the retinal vasculature from the area of the optic nerve head. Altering this VEGF expression is associated with abnormalities in retinal vascular outgrowth (51
52
53
54)
. Thus, the persistence of hyaloid vessels in the vitreous may affect hypoxia-induced VEGF expression in neuroglial cells, leading to a reduced retinal VEGF expression and eventually resulting in an abnormal outgrowth of the retinal vasculature. Consistent with this hypothesis, the delayed regression of hyaloid vessels and abnormal outgrowth of retinal vessels in Col18a1–/– mice are associated with reduced retinal VEGF expression (3)
.
A delayed hyaloid vessel regression has been noted in mutant mice with alterations in genes that are involved in the Wnt signaling pathway. For example, Lrp5 null mice (Lrp5 functioning as a Wnt coreceptor) show a significant delay in hyaloid vessel regression (55)
, similar to what is seen in patients with osteoporosis-pseudoglioma syndrome with recessive loss-of-function mutations in LRP5 (56)
. Loss-of-function mutations in the gene encoding the Wnt receptor frizzled-4 result in a delayed regression of hyaloid vessels in mice and humans (57
58
59)
. Moreover, mutations in the gene encoding Norrin, a ligand for frizzled-4 that activates the Wnt pathway, cause a delayed regression of hyaloid vessels (in addition to more complex vascular ocular abnormalities) (59
60
61)
. These findings are intriguing because collagen XVIII/endostatin has been implicated in Wnt signaling. The N-terminal domain of the longest collagen XVIII isoform contains a frizzled-like motif, and the C-terminal endostatin domain has been reported to induce ß-catenin degradation (62
, 63)
. Although these specific domains of collagen XVIII have not been shown to affect hyaloid vessel regression directly, altered Wnt signaling may contribute to the delayed hyaloid vessel regression in Col18a1 null mice.
It is not known whether the lack of collagen XVIII/endostatin affects primarily the apoptosis-inducing function of macrophages in the process of hyaloid vessel regression (50)
or the intrinsic properties of the hyaloid vasculature. In favor of the latter, recent experiments have suggested intrinsic differences of the vasculature of Col18a1 null mice when compared to wild-type mice. Endothelial cells isolated from Col18a1–/– mice showed an increased ability to adhere to fibronectin in vitro when compared to wild-type endothelial cells; increased outgrowth of microvessels from aortic explants of Col18a1–/– mice has been observed as well (17)
. In this study, the increased microvessel outgrowth in aortic explants from Col18a1–/– mice was reduced to the wild-type level by the addition of low levels of recombinant endostatin during the culture period. Notably, the net increase in microvessel outgrowth seemed to be a result of a reduced regression in the dynamic process of elongation and subsequent regression of newly formed microvessels (17)
. These findings suggest that collagen XVIII/endostatin may regulate interactions between endothelial cells and the underlying BM. Although the mechanism of hyaloid vessel regression is not well understood, it is likely that intrinsic differences of adhesive properties between endothelial cells and the fibronectin-containing BMs of hyaloid vessels contribute to the delayed regression of these vessels in mutant mice.
That collagen XVIII/endostatin is important for vascular function in some tissues is further supported by recent data showing increased vascular permeability in the skin of mice lacking even a single Col18a1 allele (64)
. This increase in vascular permeability in mutant mice was not suppressed by increasing serum endostatin to wild-type levels.
In contrast, administration of endostatin did inhibit vascular permeability in wild-type mice if induced through high-level expression of VEGF, such as in experiments using a viral vector system to express endostatin intraocularly and assess its effect on VEGF-induced retinal vascular permeability (65)
. The results of these experiments are of clinical significance, since endostatin could be used therapeutically to target the VEGF-induced increase in vascular retinal permeability in diseases such as diabetic retinopathy. Moreover, high serum levels of endostatin (via intravenous injection of adenoviral vectors expressing secretable endostatin) reduced laser-induced choroidal neovascularization, a model for neovascularization as seen in age-related macular degeneration (66
, 67)
.
The result of permeability assays implies altered cell-matrix interactions and vascular properties due to the lack of collagen XVIII. However, the lack of clinical edema or vascular abnormalities in Knobloch syndrome patients and Col18a1 null mice shows that collagen XVIII/endostatin is not a critical regulator of vascular permeability in vivo, but probably only one contributory component. Thus, there may be functional or structural abnormalities in endothelia or epithelia of mice lacking collagen XVIII/endostatin that are not obvious and become apparent only when tissue function is challenged in a certain assay system or is influenced by additional genetic alterations. When the original Col18a1 null mouse strain (which had been backcrossed into a C57BL/6J genetic background) was crossed into a different genetic background (C57BL/6JolaHsd strain), a predisposition for the development of hydrocephalus was evolving that was not observed in the original mutant strain (21)
. The choroid plexus epithelium in that new mutant strain showed marked structural alterations, suggestive of increased secretion of cerebrospinal fluid. Thus, lack of collagen XVIII/endostatin does not cause an overt hydrocephalus phenotype in all strains of mice, but is a predisposing factor to this phenotype, and the phenotype only becomes apparent in conjunction with additional genetic influences.
|
AGE-DEPENDENT LOSS OF VISION IN Col18a1–/– MICE IS ASSOCIATED WITH ABNORMALITIES OF THE RETINAL PIGMENT EPITHELIUM (RPE) AND FORMATION OF BASAL LAMINAR DEPOSITS |
|---|
|
TOPABSTRACTINTRODUCTIONCOLLAGEN XVIII/ENDOSTATIN GENE...ORGANIZATION OF COLLAGEN XVIII...ABNORMAL VASCULATURE DUE TO...AGE-DEPENDENT LOSS OF VISION...EVIDENCE FOR A FUNCTIONALLY...ABNORMALITIES OF THE IRIS...MIGRATION OF PIGMENTED IRIS...HAS COLLAGEN XVIII/ENDOSTATIN...SUMMARYREFERENCES |
|---|
. These deposits show striking ultrastructural similarities to basal laminar deposits, as seen in early stages of human age-related macular degeneration (ARMD) (68)
, with amorphous electron-dense material and membranous debris forming between the RPE and Bruch’s membrane. Deposits were found throughout the sub-RPE space of the eye, including the peripheral retinal region. Basal laminar deposits in affected human eyes consist at least in part of excess BM-like material, most likely produced by the RPE (69)
. Sub-RPE deposits in mutant mice do contain BM components, such as collagen IV. Based on these findings, Col18a1–/– mice may serve as a model for studying the mechanisms that underlie pathological basal laminar deposit formation in early ARMD. However, Col18a1–/– mice do not show major features of advanced ARMD, since no typical drusen are formed and no choroidal neovascularization is seen. Thus, these mice are not a mouse model for all features of ARMD, but for basal laminar deposit formation (although it cannot be excluded that features of advanced ARMD may develop in Col18a1 null mice after a longer time).
In association with pathological sub-RPE deposit formation, the RPE in aged mutant mice shows functional and morphological abnormalities. RPE retinoid metabolism is affected with reduced retinyl esters and RPE65 protein (essential for the formation of 11-cis retinal) (18)
. These changes are associated with retinal abnormalities in aged mutant mice with reduced retinal rhodopsin levels. This explains the decreased ERG amplitudes and loss of visual sensitivity, since photoreceptors depend on a proper function of the RPE (supplying 11-cis retinal to photoreceptors and performing the daily phagocytosis of the shed distal ends of the photoreceptor outer segments). A reduced interdigitation of the apical villi of the RPE with abnormally bent photoreceptor outer segments and widened RPE basal infoldings were found in aged Col18a1–/– mice (18)
, similar to what has been observed in aged human eyes with early ARMD (69)
. Retinal disorganization in aged mutant mice was pronounced at areas with increased retinal glial fibrillary acidic protein (GFAP) expression. Such overexpression of GFAP in retinal Müller cells is often found as a consequence of abnormalities of the retina (70)
. In summary, the changes observed in the retina of aged mutant mice show similarities to changes observed in ARMD in humans (70)
. These changes are likely consequences of RPE defects due to an altered Bruch’s membrane in Col18a1–/– mice (Fig. 3
).
|
Although an abnormal pattern of the retinal vasculature was observed in young Col18a1–/– mice, it is unlikely that the abnormalities seen in the RPE and retina of aged mutant mice are consequences of these vascular abnormalities. First, the abnormally patterned retinal vessels are located in the layers of the retina closest to the vitreous. Second, fluorescent angiography experiments demonstrated no obvious deficiency in the perfusion of the retina. The data suggest that the extracellular matrix (ECM) component collagen XVIII/endostatin is essential for the maintenance of the RPE and imply an important role for this collagen for Bruch’s membrane function.
Consistent with the observation that collagen XVIII is essential for RPE function, fundoscopic examinations in Knobloch syndrome patients revealed RPE abnormalities (47
, 71)
. However, the question remains as to why lack of a BM component leads to the formation of basal laminar-like deposits. These deposits form because of an excess production of BM-like material by the RPE or because of a reduced clearance of such material. It is possible that the absence of collagen XVIII/endostatin might cause subtle functional and structural changes of the highly complex Bruch’s membrane, resulting in altered RPE cell function. Experiments with mice suggest that direct RPE cell damage (as in experiments with photochemical injury of the RPE; ref 73
) can cause abnormal basal laminar-like deposit formation. Perhaps a structurally altered Bruch’s membrane (as in Col18a1–/– mice) (18)
results in cell stress and sub-RPE deposit formation via similar mechanisms.
|
EVIDENCE FOR A FUNCTIONALLY ALTERED BM DUE TO THE LACK OF COLLAGEN XVIII/ENDOSTATIN |
|---|
|
TOPABSTRACTINTRODUCTIONCOLLAGEN XVIII/ENDOSTATIN GENE...ORGANIZATION OF COLLAGEN XVIII...ABNORMAL VASCULATURE DUE TO...AGE-DEPENDENT LOSS OF VISION...EVIDENCE FOR A FUNCTIONALLY...ABNORMALITIES OF THE IRIS...MIGRATION OF PIGMENTED IRIS...HAS COLLAGEN XVIII/ENDOSTATIN...SUMMARYREFERENCES |
|---|
). In these models advanced age and hypercholesterolemia (induced by an atherogenic diet or by genetic alterations) predisposed to basal laminar deposit formation between the RPE and Bruch’s membrane. Significant basal laminar deposits accumulated in mice fed an atherogenic diet and subsequently exposed to laser photochemical injury of the RPE (73)
. Hypercholesterolemia is believed to lead to a thickening of Bruch’s membrane and accumulation of lipid-rich membranous debris within Bruch’s membrane, possibly compromising RPE metabolism. Thus, the combination of laser exposure and Bruch’s membrane alterations due to hypercholesterolemia may induce RPE dysfunction and basal laminar deposit formation.
|
Examples of mouse models with hypercholesteremia that develop massive age-dependent accumulation of basal laminar-like deposits include ApoB100 transgenic mice or Apo(*)E3-Leiden transgenic mice fed an atherogenic diet (74
, 75)
. Other mouse models for basal laminar deposit formation have been described using photochemical injury of the RPE partly in combination with an atherogenic diet or the addition of estradiol (73
, 76
77
78)
(Table 2)
.
These models do not provide a unifying hypothesis as to what the pathogenetic mechanisms are in basal laminar deposit formation in early ARMD in humans. However, they suggest that the observed deposits in Col18a1–/– mice may not be a specific consequence of the lack of this collagen in Bruch’s membrane. Instead, they may represent relatively unspecific consequences of RPE dysfunction.
Col18a1–/– mice represent the first mouse model for basal laminar deposit formation due to deficiency of a single component of Bruch’s membrane (18)
. Studies of these mice demonstrate the importance of a proper Bruch’s membrane for RPE function and show that BM defects can be a primary pathogenetic cause for basal laminar deposit formation.
This is supported by recent reports of the identification of mutations in genes of several different BM components that result in a progressive loss of RPE function and basal laminar deposit or drusen formation in humans. These gene mutations are the cause of clinically distinguishable but similar ocular diseases. Missense mutations in the gene encoding the ECM component fibulin-5 have been identified in patients with ARMD (79)
. Since fibulin-5 is important for the formation of elastic fibers, it is possible that missense mutations in fibulin-5 may affect the assembly or remodeling of elastic fibers in Bruch’s membrane. Alternatively, interaction of the RPE with Bruch’s membrane may be altered through these missense mutations, as fibulin-5 is a ligand for several integrins (80
, 81)
. Mutations in the gene encoding the ECM component fibulin-3 have been associated with other human ocular diseases that are characterized by abnormal deposits under the RPE (central confluent yellow-white drusen), namely, Malattia Leventinese and Doyne honeycomb retinal dystrophy (82)
. Some initial data suggest an association between mutations in the gene encoding fibulin-6 and ARMD (83)
. Mutations in a recently identified short-chain collagen gene, CTRP5, have been found in patients with autosomal dominant late-onset retinal degeneration, also characterized by abnormal deposits under the RPE (84)
. In all these ocular diseases, abnormal deposits accumulate within the RPE-Bruch’s membrane-choroid complex due to mutations in distinct components of the ECM (Table 2)
. The pathogenetic mechanisms in these diseases remain unknown, but it is tempting to speculate that mutations in the genes for fibulin-3, fibulin-5, fibulin-6, CTRP-5, or collagen XVIII may all have a similar net result, namely, an altered BM structure that affects cell-matrix interactions. It may well be that these ECM proteins are part of the same ultrastructure or protein complex and that mutations in any of these components result in similar defects and abnormal deposit formation. For example, it is known that some fibulins (at least fibulin-1 and fibulin-2) can bind with high affinity to the carboxyl domain of collagen XVIII (32)
.
In conclusion, it is likely that collagen XVIII affects BM function in several ways and that a lack of these functions is apparent in specific tissues that are frequently affected through mutations in ECM components, such as in Bruch’s membrane. In other tissues, lack of collagen XVIII does not result in an obvious pathologic change in BM function. Abnormally thickened BMs have been described in a variety of tissues in the Col18a1–/– mouse strain, which developed a predisposition for hydrocephalus formation (21)
without affecting most of the adjacent epithelial or endothelial cells in a phenotypically significant way.
|
ABNORMALITIES OF THE IRIS AND CILIARY BODY IN Col18a1–/– MICE |
|---|
|
TOPABSTRACTINTRODUCTIONCOLLAGEN XVIII/ENDOSTATIN GENE...ORGANIZATION OF COLLAGEN XVIII...ABNORMAL VASCULATURE DUE TO...AGE-DEPENDENT LOSS OF VISION...EVIDENCE FOR A FUNCTIONALLY...ABNORMALITIES OF THE IRIS...MIGRATION OF PIGMENTED IRIS...HAS COLLAGEN XVIII/ENDOSTATIN...SUMMARYREFERENCES |
|---|
, 85)
. Thus, collagen XVIII may be important not only for RPE function, but may have a critical role in the proper functioning of the iris pigment epithelium (IPE). This possibility is likely based on the similar properties of RPE and IPE cells and their underlying BMs. Cultured IPE cells can acquire properties of RPE cells (such as photoreceptor phagocytosis or retinol metabolism), and both cell types can dedifferentiate into precursor cells. Depending on the appropriate stimuli, these precursor cells can redifferentiate into RPE, IPE, lens epithelium, or neural cells (reviewed in ref 86
).
The iris stroma is separated from the single-layered anterior IPE by a BM. Anterior IPE cells are connected at their apices with the posterior IPE cells via junctional complexes. The vitreal surface of the posterior IPE contains a second iris BM, which is continuous with the posterior ciliary body epithelial BM. These BMs all contain collagen XVIII (3
, 19
, 20)
. Mice lacking collagen XVIII have defects in the iris with rupture of the posterior IPE cell layer and pigment dispersion, or separation between the two IPE cell layers (19
, 20)
. The iris abnormalities in mutant mice are similar to those observed in patients with pigment dispersion syndrome. Thickening of the stromal iris BM and a flattening of the nonpigmented ciliary body epithelium were noted in mutant mice as well. Thus, the absence of collagen XVIII alters the properties of iris and ciliary body BMs and results in defects in the posterior epithelial cell layer of these ocular tissues. The defects are possibly due to abnormal cell-matrix interactions and changes in cytoskeletal organization in the epithelial cells.
Several in vitro experiments indicated an effect of recombinant endostatin on cell signaling and cytoskeletal assembly (40
, 87)
. The actin stress fiber network in endothelial cells was dissociated in response to endostatin treatment in one study (88)
. It has been suggested that endostatin treatment may induce tyrosine phosphorylation of focal adhesion kinase and paxillin, and under some conditions even promote the formation of focal adhesions and actin stress fibers (89)
. A role for recombinant endostatin in cell adhesion and cytoskeletal assembly is suggested by its anti-migratory effect on endothelial cells in vitro (90)
. Changes in the ß-catenin pathway have been reported in response to endostatin treatment as well (62
, 63)
, affecting cell-cell adhesion and cell motility. Whereas ß-catenin is tyrosine phosphorylated in endothelial cells in response to angiogenic growth factors (such as VEGF or FGF-2), resulting in increased cell motility, no phosphorylation was observed when treating cells with endostatin in addition to such growth factors (reviewed in ref 91
). These in vitro studies indicate a role for recombinant endostatin in cytoskeletal organization and cell-matrix interactions. Based on these studies, one may speculate that lack of the endostatin domain in Col18a1–/– mice would affect cell-matrix interactions. However, the studies were done using mostly endothelial cell lines, and it remains to be shown whether epithelial defects in Col18a1 null mice can be explained by cytoskeletal abnormalities. Nor is it known whether the ocular epithelial cell abnormalities observed in Col18a1–/– mice are caused by the absence of collagen XVIII or are due to a lack of endostatin only.
|
MIGRATION OF PIGMENTED IRIS "CLUMP" CELLS TOWARD THE RETINA |
|---|
|
TOPABSTRACTINTRODUCTIONCOLLAGEN XVIII/ENDOSTATIN GENE...ORGANIZATION OF COLLAGEN XVIII...ABNORMAL VASCULATURE DUE TO...AGE-DEPENDENT LOSS OF VISION...EVIDENCE FOR A FUNCTIONALLY...ABNORMALITIES OF THE IRIS...MIGRATION OF PIGMENTED IRIS...HAS COLLAGEN XVIII/ENDOSTATIN...SUMMARYREFERENCES |
|---|
). These cells have the ultrastructural appearance of so-called iris "clump cells of Koganei," with variable size and form of pigment granules and plasma membrane protrusions (92
, 93)
. Their function is unknown, but they express the murine macrophage marker F4/80 (19)
. These pigmented cells originate from the iris stroma, migrate toward the retina, and are able to penetrate the ILM. They accumulate particularly at regions of the vitreal-retinal interface, where the retina shows signs of severe retinal disorganization and high expression levels of GFAP in mutant mice (18)
. One may speculate that the combination of structural abnormalities in the iris and local abnormalities in the retina of aged Col18a1 null mice lead to the directed migration of these macrophage-like cells toward areas of retinal defects. No migrating iris clump cells have been detected in the retina or vitreous of young Col18a1 null mice. Such migrating cells have been detected only in aged mutant mice with retinal abnormalities. Thus, clump cells may function in the repair or clearance of retinal defects resulting from age-dependent increase in RPE dysfunction in these mice.
|
|
HAS COLLAGEN XVIII/ENDOSTATIN SPECIFIC FUNCTIONS FOR OCULAR EPITHELIA? |
|---|
|
TOPABSTRACTINTRODUCTIONCOLLAGEN XVIII/ENDOSTATIN GENE...ORGANIZATION OF COLLAGEN XVIII...ABNORMAL VASCULATURE DUE TO...AGE-DEPENDENT LOSS OF VISION...EVIDENCE FOR A FUNCTIONALLY...ABNORMALITIES OF THE IRIS...MIGRATION OF PIGMENTED IRIS...HAS COLLAGEN XVIII/ENDOSTATIN...SUMMARYREFERENCES |
|---|
In contrast to the RPE defects seen in aged mutant mice, abnormalities of the posterior IPE cell layer were observed already in 1-wk-old Col18a1–/– mice. Thus, it is likely that collagen XVIII plays an essential role for the integrity of the posterior IPE cell layer and that this role cannot be compensated for by other components of either the iris BMs or the IPE. The striking cellular defect seen in the posterior IPE was not detected in the RPE or the ciliary body epithelium, even in the oldest mutant mice examined. Therefore, the requirement of epithelial cells for collagen XVIII in the BM differs even within the various ocular epithelial cell types. A detailed comparison of the cytoskeletal organization and cell adhesion properties between these different ocular and nonocular epithelial cells in mutant mice may help answer the question of whether collagen XVIII has distinct functions in these ocular epithelia.
|
SUMMARY |
|---|
|
TOPABSTRACTINTRODUCTIONCOLLAGEN XVIII/ENDOSTATIN GENE...ORGANIZATION OF COLLAGEN XVIII...ABNORMAL VASCULATURE DUE TO...AGE-DEPENDENT LOSS OF VISION...EVIDENCE FOR A FUNCTIONALLY...ABNORMALITIES OF THE IRIS...MIGRATION OF PIGMENTED IRIS...HAS COLLAGEN XVIII/ENDOSTATIN...SUMMARYREFERENCES |
|---|
It is conceivable that distinct disease mechanisms underlie the different abnormalities observed in Col18a1 null mice or in Knobloch syndrome patients. For example, neuronal cell migration defects, the altered pattern of the retinal vasculature, iris defects, hydrocephalus predisposition, delayed regression of hyaloid vessels, or the increased vascular permeability may result from the lack of activities mediated through the endostatin domain. The absence of the promigratory activity of endostatin-oligomers may change the astrocytic template that precedes retinal vascular outgrowth and so may lead to the abnormal retinal vasculature in Col18a1–/– mice. The delayed hyaloid vessel regression in these mutant mice may be a consequence of altered Wnt signaling, apoptosis, or endothelial cell adhesion properties in the absence of endostatin. Defects in the posterior epithelium of the iris and ciliary body may result from alterations in epithelial cytoskeletal organization as a consequence of the lack of the endostatin domain as well. All these distinct defects occur with early onset in mutant mice.
In contrast, RPE defects and basal laminar deposit formation occur only gradually and at an advanced age in mice. It is certainly possible that these defects are not a consequence of the lack of the endostatin domain, but rather a consequence of the lack of collagen XVIII in BMs. After proteolytic removal of endostatin from full-length collagen XVIII molecules, the remaining molecule is still part of BM networks (18)
, since collagen XVIII is anchored into perlecan-containing BM molecular networks not only through the endostatin domain. The role of collagen XVIII in the BM is not known, but the delayed onset of RPE defects in Col18a1 null mice points to the possibility that subtle structural BM changes may occur over time due to the absence of this collagen, eventually resulting in altered cell-matrix interactions and striking defects in nonregenerating epithelia, as in the RPE.
It is likely that lack of collagen XVIII affects the interaction of vascular BMs with endothelial cells, since an increased vascular permeability was detected in Col18a1 mutant mice and an increased angiogenic sprouting in in vitro assays using aortas from these mice.
Future studies in which the phenotypes of mutant mice that have mutations in specific domains of collagen XVIII are compared with the defects observed in Col18a1 null mice may help answer the question of which pathological process is due to the abnormal function of which domain within collagen XVIII. Furthermore, mutant mice that lack collagen XVIII/endostatin and other factors involved in the regulation of angiogenesis may reveal an altered angiogenic response in tissues that are otherwise not phenotypically affected in Col18a1 null mice. Such studies may help determine the extent of the role of collagen XVIII/endostatin as a physiological regulator of angiogenesis.
Received for publication June 4, 2004. Accepted for publication December 22, 2004.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONCOLLAGEN XVIII/ENDOSTATIN GENE...ORGANIZATION OF COLLAGEN XVIII...ABNORMAL VASCULATURE DUE TO...AGE-DEPENDENT LOSS OF VISION...EVIDENCE FOR A FUNCTIONALLY...ABNORMALITIES OF THE IRIS...MIGRATION OF PIGMENTED IRIS...HAS COLLAGEN XVIII/ENDOSTATIN...SUMMARYREFERENCES |
|---|
1(XVIII), a collagen chain with frequent interruptions in the collagenous sequence, a distinct tissue distribution, and homology with type XV collagen. Proc. Natl. Acad. Sci. USA 91,4234-42381 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. 269,13929-139351(XVIII) collagen gene to mouse chromosome 10 and human chromosome 21. Genomics 19,494-499[CrossRef][Medline]
