|
|
|
Vol. 14, No. 17, pp. 2123-2133, September 1, 2000
1 Department of Medicine and Lung Biology Center, and 2 Department of Anatomy, University of California, San Francisco, California 94143 USA
The matrix metalloproteinase (MMP) family of extracellular
proteinases regulates development and physiologic events. Genetic analyses using transgenic mice that have gain and loss of function of
MMPs or of their endogenous inhibitors, the TIMPs, and pharmacogenetic studies with chemical inhibitors have begun to elucidate the roles that
they play. It is now clear that these enzymes are important for cell
migration, invasion, proliferation, and apoptosis. They regulate many
developmental processes, including branching morphogenesis, angiogenesis, wound healing, and extracellular matrix degradation.
The matrix metalloproteinases (MMPs) are a family of extracellular
matrix (ECM)-degrading enzymes that shares common
functional domains and activation mechanisms
(Sternlicht et al. 2000
![]()
Introduction
Top
Introduction
MMPs and cell migration
ECM-dependent cell...
MMPs and tissue morphogenesis
MMPs and modulation of...
MMP function during...
MMPs and wound healing
MMPs in mammary development
MMPs in bone development
Conclusions
References
). These are Ca2+- and
Zn2+-dependent endopeptidases that are active at neutral pH.
They are synthesized as secreted or transmembrane proenzymes and
processed to an active form by the removal of an amino-terminal
propeptide. The propeptide is thought to keep the enzyme in latent form
by the interaction of a cysteine residue in this peptide with the zinc
moiety in the enzyme active site. Disruption of this interaction triggers the cysteine switch mechanism and results in activation of the
enzyme. MMPs can be activated by chaotropic agents or by cleavage of
the propeptide by members of the MMP family or by other proteases. They
are inhibited by a family of tissue inhibitors of metalloproteinases,
the TIMPs. As a family, MMPs degrade most components of the ECM. There
are now >20 members of the MMP family. There are several distinct
subgroups based on preferential substrates or similar structural
domains: Collagenases that are active against fibrillar collagen,
gelatinases that have high activity against denatured collagens,
stromelysins that degrade noncollagen components of the ECM,
membrane-type MMPs (MT-MMPs) that are transmembrane molecules, and
other less characterized members (Fig. 1; Table 1).

View larger version (37K):
[in a new window]
Figure 1.
Structural domains of the matrix metalloproteinases
(modified from Sternlicht et al. 2000
). Diagrams of the domain
structure of the different subgroups of the matrix metalloproteinase
family. The prodomain contains a highly conserved sequence with an
unpaired cysteine sulfhydryl group whose interaction with the active
site zinc maintains the enzyme in latent form (the cysteine switch).
The prodomain of some MMPs contains a recognition motif for furin-like
enzymes, whose cleavage results in activation of the MMP. The catalytic
domain of the gelatinases (MMP-2 and MMP-9) contains three
gelatin-binding fibronectin type II repeats. The hemopexin/vitronectin
domain is folded into a four-bladed propeller structure. (A)
MMPs with minimal domain. (B) MMPs with hemopexin/vitronectin
domain.
Table 1.
The family
of MMPs
Because MMPs can degrade ECM molecules, their main function has been
presumed to be remodeling of the ECM. They are thought to play
important roles during embryonic development, as ECM remodeling is a
critical component of tissue growth and morphogenesis. In fact, the
discovery of MMPs was based on the observation that during amphibian
metamorphosis, a collagenolytic activity has to be present to digest
the collagens in tadpole tails (Gross and Lapiere 1962
). The activity
of MMPs during embryonic development may extend to more than the
removal of unwanted ECM molecules, however. It is now clear that MMPs
not only remodel the ECM, but also influence many cellular functions.
MMP activity may be required during development and normal physiology
in several ways: (1) to degrade ECM molecules and allow cell migration;
(2) to alter the ECM micro-environment and result in alteration in
cellular behavior; (3) to modulate the activity of biologically active molecules by direct cleavage, release from bound stores, or the modulating of the activity of their inhibitors (Fig
2). During tissue morphogenesis any
number of these activities may contribute to the role that each MMP
plays in a developmental process. Although insights into the activities
of MMPs have emerged from in vitro studies, genetic and pharmacogenetic
studies now indicate that MMPs do have important influence on many
cellular functions. Two general approaches have been employed to
identify the roles of MMPs during mammalian development: (1) general or
tissue-specific expression of a transgene encoding an MMP or an MMP
inhibitor (TIMP), and (2) generating null mutations in an MMP gene or
TIMP gene using targeted mutagenesis. These approaches have given
insights into the roles of several MMPs in development and normal
physiology. The range of developmental effects seen in these function
perturbation studies suggests that these enzymes do indeed participate
as essential effectors of developmental processes in vivo.
|
|
| |
MMPs and cell migration |
|---|
|
|
|---|
It is evident that the ECM presents a barrier to cell migration.
Cells that adhere to the ECM have to change from an adhesive phenotype
to a migratory phenotype before they can move. This change requires
several components: activation of cytoskeletal motor function to
provide cell movement, modulation of adhesive sites and cell-surface
adhesive molecules to provide traction, clearing of ECM to break down
physical barriers, and the presence of chemoattractants to guide
migration. MMPs may potentially regulate any of these processes. Many
in vitro studies have shown that MMPs can degrade a variety of ECM
substrates, including collagens and noncollagenous molecules,
suggesting that they can function as ECM-clearing enzymes during cell
migration (Sternlicht et al. 2000
). MMPs can also modulate the function
of many biologically active molecules (see below) that participate in
chemoattraction and migration. A role for MMPs during development is
suggested by the necessity of cell migration, sometime over long
distances, during tissue morphogenesis. This is evident as early as
implantation in the extraembryonic tissues and during gastrulation in
the embryos.
Studies using assays of cell migration through ECM barriers in culture
have implicated MMPs in the migration of a variety of epithelial,
mesenchymal, and neuronal cells either through the ECM or on specific
ECM substrates. At the implantation site, trophoblasts must invade into
the maternal decidua. These cells express high levels of MMP-9
(gelatinase B). In culture a neutralizing antibody against MMP-9
inhibits trophoblasts from invading and degrading ECM (Behrendtsen and
Werb 1997
). During angiogenesis, endothelial cells migrate into the
surrounding ECM. In culture, the migration of endothelial cells through
collagen or fibrin gels is impaired by MMP inhibitors (Fisher et al.
1994
; Hiraoka et al. 1998
). During bone remodeling, osteoclasts are
recruited to the bone surfaces, and their migration may also depend on
MMPs. In culture, MMP inhibitors curb the migration of purified
osteoclasts through a collagen matrix (Sato et al. 1998
). During the
development of the nervous system, neurons may extend processes over
long distances to form connections. In a model of neuronal
differentiation, the neuroblastoma cell line SKSNBE expresses neuronal
markers and establishes extensive neurite outgrowths in response to
retinoic acid. MMP-9 is induced and found localized in the neurites,
suggesting that it participates in neurite growth
(Chambaut-Guérin et al. 2000
). This is supported by another study
showing that oligodendrocytes in culture expresses active
metalloproteinases at the tip of their processes and that a
neutralizing antibody against MMP-9 inhibits these outgrowths (Oh et
al. 1999
). Interestingly, oligodendrocytes from MMP-9-deficient mice
fail to produce outgrowths in culture.
Epithelial morphogenesis depends on specific movements of epithelial
cells. In culture, keratinocyte migration on collagen-1 requires
specific cleavage of the collagen-1 molecule by a collagenase. MMP
inhibitors curtail keratinocyte migration, and cells not expressing MMP-1 (collagenase-1) do not migrate on this substrate. In addition, keratinocytes do not migrate on a mutant type I collagen that lacks the
collagenase cleavage site (Pilcher et al. 1997
). In another system,
breast epithelial cell migration on laminin-5 correlates with
expression of MMP-14 (MT1-MMP) and MMP-2 (gelatinase A), and
anti-sense oligodeoxynucleotides against MT1-MMP reduced their
migration over this substrate (Koshikawa et al. 2000
).
Organ cultures also show the requirement of MMPs in cell migration in
systems that more closely approximate in vivo development. In early
long bone development, (pre)osteoclastic cells invade into the
cartilage anlage to initiate the process of endochondral ossification.
These cells express high levels of MMP-9, and their migration into the
cartilage matrix of skeletal elements in organ culture is limited by
MMP inhibitors (Blavier and Delaissé 1995
). In an explant culture
model, MMP inhibitors dramatically alter mandibular morphogenesis (Chin
and Werb 1997
). In the absence of MMP activity, tongue development is
severely affected. There is impaired migration and proliferation of
myoblasts into the tongue buds, the oral sulci fail to form, and fusion
of the medial sulcus does not occur. Failure of the fusion of the
medical sulcus may be secondary to failure of epithelial cells to
migrate away from the medial edges, and failure of the formation of the
oral sulcus may result from impaired epithelial invagination, which in
turn requires epithelial cell migration into the underlying matrix. In
this model, there is a delay in the maturation and fusion of the
component segments of Meckel's cartilage. This may be due to failure
of the remodeling of the perichondrium that would allow cartilage
elements to expand into the surrounding mesenchyme and to fuse.
Interestingly, mandibles from mouse embryos lacking the EGF receptor
express much lower amounts of MMPs and show an identical developmental
defect in culture (Miettinen et al. 1999
).
| |
ECM-dependent cell proliferation and apoptosis |
|---|
|
|
|---|
Interactions between ECM components and cell surface molecules
regulate cell behavior. In a changing environment, the ability of cells
to proliferate, survive, or differentiate changes as well. It is not
surprising then that MMPs can alter cell behavior through their action
on the ECM. Studies on the roles of MMPs on cell behavior in culture
are few but have shown effects on cell proliferation, survival,
apoptosis, differentiation, and organization. Inhibition of MMPs either
by a chemical inhibitor or a neutralizing antibody to MMP-2 reduces the
mitogenic response of cultured vascular smooth muscle cells to platelet
derived growth factor (PDGF) (Uzui et al. 2000
). Cultured airway smooth
muscle cells also depends on MMP-2 for the proliferative response to a
mitogen (Johnson and Knox 1999
). These cells proliferate and secrete
MMP-2 in response to treatment with either serum or PDGF, and an MMP
inhibitor suppresses the proliferative response. Similarly, cultured
human dermal microvascular endothelial cells (HDMEC) increase ECM
deposition and suppress cell proliferation when MMP activities are
inhibited (Kräling et al. 1999
). In this case, MMP activity may
be required to generate a proliferative signal. MMPs have also been
shown to cause cell death. Proteinases or inappropriate ECM molecules
induce apoptosis of mammary epithelial cells in culture presumably
through altered signaling from integrins (Boudreau et al. 1996
).
ECM-degrading proteinases are induced in mammary cells in response to a
fibronectin fragment (FN120), and inhibition of MMP activity rescues
cell apoptosis (Schedin et al. 2000
). In addition, MMP inhibitors also
suppress smooth muscle cell proliferation and induce cell apoptosis in
hypertrophied rat pulmonary arteries in organ culture (Cowan et al.
2000
). Thus, MMPs affect cell survival and proliferation both
positively and negatively by regulating survival signals generated by
specific adhesive events. These differences in the effects of MMPs may reflect the differences in MMP substrates involved in each response.
| |
MMPs and tissue morphogenesis |
|---|
|
|
|---|
A particular action of MMPs of specific relevance to tissue
morphogenesis is their ability to affect cell organization. When adipocytes are cultured on basement membrane, they migrate and organize
into large multicellular clusters. These cells secrete MMP-2, and
inhibition of MMP activity results in inhibition of cell migration and
organization into three-dimensional structures (Brown et al. 1997
). In
an in vitro model of pancreatic islet development, embryonic pancreatic
epithelial cells cultured in collagen gels differentiate and organize
into clusters with characteristics of islets of Langerhans. Beta cells
are located in the center and non-beta cells in the periphery. MMP-2
activity is induced, and inhibition of MMP activity by a chemical
inhibitor abolishes islet morphogenesis without affecting endocrine
cell differentiation (Miralles et al. 1998
). Cells differentiate, but
do not migrate and associate into islets.
Another ECM-dependent cell organization process is seen during tubular
and branching morphogenesis, during angiogenesis, and during the
development of many epithelial structures. Regulated and localized ECM
remodeling may be required during tubular and branching morphogenesis
because changes in cell proliferation, motility, shape, and formation
of cellular contacts necessary for this process require precise
interactions between cells and the ECM. Formation of tubular structures
by endothelial and epithelial cells in three-dimensional culture shows
varied dependence on MMPs. Tubule formation by HUVECs cultured on
Matrigel or collagen gels is inhibited by inhibiting MMP activity
(Schnaper et al. 1993
; Fisher et al. 1994
). Bovine aortic endothelial
cells in collagen gels undergo apoptosis unless an angiogenic factor is present, in which case they survive, proliferate, and form
capillary-like tubes. In this system, MMP inhibitor prevents apoptosis
in the absence of an angiogenic factor and enhances tube formation in response to angiogenic factor (Kuzuya et al. 1999
). Mammary epithelial cells in three-dimensional culture organize more into ductal structures than alveolar-like structures in response to a fibronectin fragment (FN120), and this process is dependent on MMP activity (Schedin et al.
2000
). Overexpression of MMP-3 (stromelysin) in mammary epithelial
cells induces epithelial-to-mesenchymal transition (Lochter et al.
1997
; Sternlicht et al. 1999).
Organ cultures have been used extensively to show the role of MMPs in
branching morphogenesis. When day 10 embryonic kidney explants are
placed in culture, the ureteric bud develops by branching morphogenesis. This is inhibited by a neutralizing antibody against MMP-9 (Lelongt et al. 1997
). MT1-MMP may also be involved in this process, as treatment of embryonic kidney cultures with anti-sense oligodeoxynucleotides against MT1-MMP also decreases branching of the
ureteric buds (Kanwar et al. 1999
). In addition, anti-sense oligodeoxynucleotides against TIMP-2 increase branching and rescue the
effect of anti-sense oligodeoxynucleotides against MT1-MMP.
In contrast to its effect in kidney development, inhibition of MMP
activity in cultured salivary glands leads to increased branching
(Fukuda et al. 1988
). The difference in the effect of MMPs may be
secondary to the mechanism of branching in the two epithelia. Kidney
epithelium branches by budding and growing into the mesenchyme,
requiring the degradation of the surrounding ECM. Salivary gland
epithelium branches by formation and stabilization of clefts composed
of collagen bundles in the epithelium concurrently with growth into the
surrounding mesenchyme. Thus, in the kidney, ECM degradation would
increase branching, but in the salivary glands enhanced ECM degradation
would decrease branching, as it would lead to destabilization of the
collagen clefts (Fig. 3).
Taken together these data suggest that MMPs play several distinct roles: altering ECM, permitting invasion, selection of branch points, regulating three-dimension organization, regulating epithelial-to-mesenchymal transition. Clearly these complex events suggest that multiple targets for MMP action must exist.
| |
MMPs and modulation of biologically active molecules |
|---|
|
|
|---|
A recurring paradigm in biologically active molecules is the
generation of fragments with new activities by proteolytic cleavage. Several potent angiogenic inhibitors are produced by proteolytic cleavage: Angiogstatin is a fragment of plasminogen produced by cleavage in vitro by a number of MMPs, including MMP-3, MMP-7, MMP-9,
and MMP-12 (macrophage metalloelastase) (O'Reilly et al. 1994
; Dong et
al. 1997
; Patterson and Sang 1997
; Lijnen et al. 1998
), and in vivo by
MMP-7 and MMP-9 (Pozzi et al. 2000
). Mice that are deficient in the
integrin alpha-1 have decreased vascularization of tumor xenografts.
This is due to increased level of circulating angiostatin generated by
increased activity of MMP-7 and MMP-9 in these mice (Pozzi et al.
2000
). Thus, increased expression of MMPs may both confer invasiveness
to the tumor cells and, paradoxically, lead to production of molecules
that limit their growth. Other angiogenic inhibitors generated by
proteolysis have also been reported. Endostatin is a fragment of
collagen XVIII with antiangiogenic activity (O'Reilly et al. 1997
),
and cleavage of antithrombin-III results in a molecule that is an
inhibitor of angiogenesis (O'Reilly et al. 1999
).
MMPs may also regulate the function of biologically active molecules by
other mechanisms: They may control the molecules' bioavailability by
releasing them from bound proteins or ECM stores, or they may regulate
the molecules' activity by proteolytically activating or inactivating
them. Thus, MMP activity can lead to both gain and loss of function of
biologically active molecules. Examples for each of these mechanisms
have been reported. MMP-1 and MMP-3 can degrade perlecan and release
bound FGF (Whitelock et al. 1996
). Cleavage of decorin by MMP-2, MMP-3,
and MMP-7 (matrilysin) releases TGF-
bound to decorin (Imai et al.
1997
). MMPs may also regulate the bioavailability of growth factors by
cleaving the proteins that bind to them. An example is the cleavage of
insulin-like growth factor-binding proteins (IGFBPs) by MMPs. IGFBPs
bind insulin-like growth factors (IGFs) with equal or higher affinity
than IGF receptors, and degradation of IGFBPs is thought to play a
major role in the regulation of IGF activity (Rajah et al. 1995
).
IGFBP-3 can be cleaved by MMP-1, MMP-2, and MMP-3, and IGFBP-5 is
cleaved by MMP-1 and MMP-2 (Fowlkes et al. 1994a
,b
, 1995
; Thrailkill et
al. 1995
; Wu et al. 1999
). Of interest, mice overexpressing TIMP-1, an
inhibitor of MMPs, have higher serum levels of IFGBP-3, lower levels of
dissociable IGF-II, and decreased signaling through the IGF receptor
(Martin et al. 1999
). MMP activity is normally found in the liver, and
liver of TIMP-1 overexpressing animals shows decreased proteolysis of
IGFBP-3. MMP-2 and MMP-9 may also activate latent TGF-
by cleaving
its latent associated peptide (LAP; Yu and Stamenkovic 2000
).
MMPs also affect the activity of other molecules through proteolytic
cleavage. IL-1
precursor can be processed to its active form by
MMP-2, MMP-3, and MMP-9 (Schönbeck et al. 1998
). However, prolonged incubation with MMPs results in IL-1
degradation and loss of its biological activity (Ito et al. 1996
). In some cases, cleavage of a molecule by the MMPs generates a more biologically active
fragment. This is true for endothelin-1, which results in a peptide
with greater vasoconstrictor effects when cleaved by MMP-2
(Fernandez-Patron et al. 1999
). Another example is MMP-8 (neutrophil
collagenase) cleavage of angiotensin-I, which generates angiotensin-II
(Diekmann and Tschesche 1994
). In other cases, although cleavage of a
molecule by an MMP occurs, the resulting modulation of activity by such
cleavage is not known, such as cleavage of substance P by MMP-9 and
that of apolipoprotein (a) by MMP-12 (Backstrom and Tökés
1995
; Edelstein et al. 1999
). MMPs may also cleave cell surface
molecules, thereby modulating their activity. An example is MMP-2 and
MMP-9 cleavage of galectin-3 which alters the carbohydrate recognition
domain of galectin-3 and reduces self-association of the galectin
molecules. Thus, cleavage by MMPs may result in significant alteration
in the signal-transducing property of this molecule (Ochieng et al.
1994
, 1998
).
MMPs may also modulate the activity of other proteinases. They may
proteolytically activate latent proteases or inactivate their
inhibitors. MMP-3 can cleave urokinase-type plasminogen activator
(u-PA), separating the receptor-binding domain from its catalytic
domain, without affecting catalytic activity (Orgel et al. 1998
; Ugwu
et al. 1998
). The biological significance of this finding is unclear,
but it is possible that removing receptor binding would result in
decreased localization of u-PA to the cell surface. MMPs may also
regulate the activity of other proteinases by degradation of their
inhibitors. MMP-1, MMP-3, and MMP-9 can cleave
1-proteinase
inhibitor and
1-antichymotrypsin (Mast et al. 1991
). MMP-1, MMP-3,
MMP-7, and MMP-9 proteolytically inactivate
1-proteinase inhibitor
(Sires et al. 1994
).
Even though it is clear from the above studies that MMPs can modulate
the activity of a diverse number of biologically active molecules,
there has been no direct evidence implicating this mechanism of MMP
action in a developmental process. One may infer the role of MMPs in
cases where the activity of a developmentally relevant molecule may be
modulated by MMPs. For example, because FGF plays important roles in
chondrocyte proliferation and differentiation, any modulation of its
activity by MMPs by releasing it from ECM stores may be an important
regulatory mechanism for cartilage development. Similarly, because
TGF-
has an important activity as a tubulogenic factor, the
ability of MMPs to regulate its activity may be essential in branching morphogenesis.
The experiments described above have reached their conclusions by modulating MMP activity in culture. A drawback in these studies is that they have mainly used an inhibitor-based approach. Most of the inhibitors used are not completely specific to MMPs. The chemical inhibitors are broad-ranged. In addition the function discerned for an MMP in the isolated context of in vitro studies may not be its true function in the complex environment of the whole animal. Analysis of transgenic mice that have gain and loss of function of MMPs or their endogeneous inhibitors have given insights into the roles that these enzymes play in several developmental processes.
| |
MMP function during implantation |
|---|
|
|
|---|
Development of the placenta starts with the invasion and migration
of trophoblast cells into the maternal tissue to establish connection
with the maternal circulation (Rinkenberger et al. 1997
). These
trophoblasts express high levels of MMP-9 (Alexander et al. 1996a
). In
vitro MMP inhibition decreases migration and degradation of ECM by
these cells. Evidence for in vivo role for MMP-9 in implantation comes
from studies of the Ets-2- and MMP-9-null mice, as well as of TIMP-1
overexpressing mice and pharmacogenetic studies using chemical
inhibitors. Ets-2 is a member of the Ets family of transcription
factors that regulate the transcription of diverse genes, particularly
MMPs. The Ets-2-deficient mice die early in embryogenesis, owing to
defective development of the placenta (Yamamoto et al. 1998
). The
embryos die by embryonic day (E) 8.5. At E7.5, there is a smaller
amount of attached trophoblastic tissue. At E6.5, the ectoplacental
cone is small, apparently because of failure of trophoblast migration.
A membrane, immunostained for laminin, a basement membrane component of
Reichert's membrane, covers the ectoplacental cone and extends into
and over the trophoblasts. There is also poor connection with the
maternal circulation. Interestingly, in trophoblasts, expression of
MMP-9, which is a transcriptional target of Ets-2, is significantly
decreased in the Ets-2-null mice. Thus, the placental phenotype in
Ets-2 deficiency may be due to insufficient MMP-9 activity in
trophoblasts. In support of this model, TIMP-1 overexpressing,
metalloproteinase inhibitor-treated mice (Alexander et al. 1996a
) and
MMP-9-null mice on some genetic backgrounds (J. Rinkenberger and Z. Werb, unpubl.) display similar placental phenotype, even though these
do not lead to lethality. This severe lethal phenotype in the Ets-2
mice may be due to the effects of Ets-2 deficiency on other gene
products. The mild phenotype of mice with MMP-9 deficiency compared
with those overexpressing TIMP-1 or those treated with chemical
inhibitors suggests that other MMPs and/or other metalloproteinases
also contribute. Neither the chemical inhibitors nor the TIMPS are
completely specific to MMPs. For example, TIMP-1 also inhibits
aggrecanase, an ADAM-TS family member, and TIMP-3 also inhibits TACE, a
shedding enzyme that is a member of the ADAM family and regulates
availability of TGF-
.
| |
MMPs and wound healing |
|---|
|
|
|---|
Healing of a skin wound requires several processes similar to
development: cell migration, ECM degradation, and tissue reorganization (Fig. 4). Keratinocytes at the edge of the wound have
to migrate to re-epithelialize the wound surface. Then the fibrin-rich
provisional matrix that is laid down following wounding must be
removed. The dermis also contributes by contracting to facilitate wound
closure. Studies in culture have shown the need for MMP-1 activity in
keratinocyte migration. In animal studies MMP activity is implicated in
keratinocyte migration and dermal contraction. Wound healing is
retarded in mice treated with an MMP inhibitor (Lund et al. 1999
), due
to decreased keratinocyte migration. Interestingly,
plasminogen-deficient mice also have a similar wound-healing defect,
suggesting that both plasmin and MMPs are important for keratinocyte
migration during wound healing (Romer et al. 1996
). In addition,
keratinocyte migration is completely inhibited in the
plasminogen-deficient mice treated with an MMP inhibitor, showing that
these two enzyme families act synergistically (Lund et al. 1999
). As
discussed above, it is difficult to implicate only MMPs in processes
that are inhibited by a chemical inhibitor. These inhibitors are not entirely specific. Direct evidence for a role of MMPs in wound healing
comes from studies of MMP-3-null mice. These studies showed that MMP-3
is important for wound contraction. The MMP-3-null mice fail to
upregulate a multicellular contractile ring of actin in dermal
fibroblasts and fail to contract the wound, and thus delay wound
healing (Bullard et al. 1999a
). Of interest, MMP-3-null dermal
fibroblasts in culture contract collagen gels significantly less that
wild-type fibroblasts, suggesting that MMP-3 expression by fibroblasts
is important in wound contraction (Bullard et al. 1999b
).
|
| |
MMPs in mammary development |
|---|
|
|
|---|
The mammary glands form during embryonic development by the budding
of an epithelial tube into the mammary fat pad and subsequent extensive growth at the terminal end bud and branching to form the
epithelial ducts (Henninghausen and Robinson 1998
). The ducts end in
terminal lobular-alveolar units that differentiate into alveoli.
These alveoli are the milk-secreting glands during lactation. The
alveoli develop during pregnancy and lactation and regress (involute) during weaning. Because of the significant amount of ECM
remodeling that has to take place during virgin development, lactation, and weaning, MMPs would be expected to play a pivotal role in mammary branching morphogenesis.
Targeted expression of an autoactivating isoform of stromelysin-1 to
the mammary gland alters development of the gland (Sympson et al. 1994
;
Witty et al. 1995
). Virgin female mice, in whom there is low level of
the transgene expression in mammary epithelia, show abnormal morphology
of the gland. The primary ducts have supernumerary branches and the
alveoli show precocious development, with
-casein expression
comparable to that seen in glands at early to middle pregnancy.
Lactating glands have high levels of transgene expression and show loss
of basement membrane integrity, with loss of laminin and collagen IV.
This results in alveoli with decreased size and shrunken lumen. During
late pregnancy, the alveolar epithelial cells in the transgenic mice
undergo unscheduled apoptosis. Interestingly, there is enhanced
cleavage of the basement membrane molecule entactin (nidogen-1) around
apoptotic cells, which parallels extracellular MMP activity. When
stromelysin-1 transgenic mice are crossed with mice overexpressing
TIMP-1 the unscheduled apoptosis of epithelial cells is extinguished
(Alexander et al. 1996b
). These studies show that the inappropriate
expression of active stromelysin-1 can lead to altered tissue
morphogenesis. Its mechanism of action is not clear, but may include
alteration in epithelial cell proliferation and morphogenesis, leading
to increased ductal branching, as well as regulation of alveolar epithelial cell differentiation and apoptosis. Indeed, increased ductal
branching has also been observed in mice with a transgene expressing
anti-sense RNA against TIMP-1 targeted to the mammary gland (Fata et
al. 1999
). Epithelial cells in these ducts showed increased
proliferation. In contrast, implantation of TIMP-1-releasing pellets in
the mammary glands results in attenuated ductal expansion and decreased
proliferation of epithelium in terminal end buds in the vicinity of the
pellets. TIMP-1-overexpressing mice also show a decrease in ductal
invasion into the mammary fat pads (M. Sciabica and Z. Werb, unpubl.).
It may be argued that overexpression of MMPs or TIMPs may lead to a response that is not physiologic. Studies of the effects of the gain and loss of function of TIMP-1 also have the problem of specificity. As discussed above, TIMP-1 also inhibits metalloproteinases other than MMPs. The only direct evidence implicating MMPs in mammary gland development comes for studies of the stromelysin-1-deficient mice. These mice have a mild abnormality in mammary gland development, but the defect is not pronounced. There is a small decrease in branching morphogenesis of the ducts, with fewer secondary and tertiary branching. These data indicate that other MMPs may have the major responsibility for ductal morphogenesis. From expression data, the most likely candidates are MMP-2 and MT1-MMP.
| |
MMPs in bone development |
|---|
|
|
|---|
The skeleton develops by one of two processes: intramembranous or
endochondral ossification (Caplan 1988
; Karsenty 1999
). The long bones
of the axial skeleton and the spine form by endochondrial ossification,
in which a cartilage template is laid down through an orderly process
of chondrocyte proliferation and maturation to form hypertrophic
cartilage. Hypertrophic cartilage is subsequently resorbed and replaced
by bone matrix following the invasion of capillaries. The resorption of
hypertrophic cartilage ECM appears to take place through the action of
chondroclasts, cells that invade into the cartilage preceding
capillaries. Following capillary invasion, osteoblasts are recruited to
form the trabecular bones, which are subsequently remodeled by
osteoclasts. Osteoblasts and osteoclasts maintain bone homeostasis in
the adults.
The flat bones of the skull form by intramembranous ossification, whereby osteoblasts differentiate and deposit bone matrix onto a cartilage primordium, which is subsequently removed. The bone matrix is also vascularized and remodeled by osteoclasts, but how this occurs is not clear. Thus, the formation of bones requires active remodeling of specialized matrices (cartilage and bone) as well as coordinated activities of many cell types (chondrocytes, endothelial cells, chondroclasts, osteoblasts, osteoclasts, and perhaps others). Recent gene targeting studies have implicated two particular MMPs in bone development: MMP-9 and MT1-MMP.
Targeted inactivation of the MMP-9 gene resulted in a very specific
defect in endochondral bone formation (Vu et al. 1998
). Mice with this
defect accumulate hypertrophic cartilage at the skeletal growth plates.
This is not due to abnormality in proliferation and maturation of
chondrocytes, but is secondary to impaired endochondral ossification.
Vascular invasion into the hypertrophic cartilage zone is impeded. This
is coupled with a delay in terminal hypertrophic chondrocyte apoptosis,
showing that hypertrophic chondrocytes do not have a fixed life span
and automatically undergo apoptosis, but their death is coupled with
capillary invasion. Only those hypertrophic chondrocytes in proximity
to the capillaries undergo cell death. There are several models for the
function of MMP-9 in endochondral bone formation: resorption of
cartilage, regulation of the availability of an angiogenic factor, or
regulation of hypertrophic chondrocyte apoptosis. MMP-9-null
hypertrophic cartilage in culture has a decrease in net release of an
angiogenic activity. The simplest model for the action of MMP-9 in this
process is that of regulating the release from hypertrophic cartilage
ECM stores of an angiogenic factor. Thus, lack of MMP-9 leads to a decrease in vascular invasion with its attendant processes.
Alternatively, MMP-9 may have other functions and the decrease in
angiogenic activity of MMP-9-null hypertrophic cartilage may be an
indirect consequence.
The relevant angiogenic factor at the growth plate is VEGF, which is
expressed by hypertrophic cartilage. Inhibition of VEGF activity by a
soluble VEGF receptor, mFlt-IgG, leads to inhibition of vascular
ingrowth into hypertrophic cartilage (Gerber et al. 1999
). The
resulting accumulation of hypertrophic cartilage is similar to, but
more severe than, that seen in the MMP-9-null animals. However, VEGF
appears to do more than regulating angiogenesis at the growth plate.
Treatment with mFlt-IgG also leads to decreased recruitment of
chondroclasts and osteoblasts. Thus, VEGF seems to be a key regulator
of the diverse processes required for organized and coordinated
endochondral ossification. MMPs may participate in this control by
directly or indirectly regulating the net balance of VEGF activity. The
defect in endochondral ossification is the only noticeable
developmental defect in these mice, pointing to a very specialized
function of MMP-9 in development.
Targeted inactivation of MT1-MMP, a transmembrane MMP, results in
several skeletal defects that are different from that in the MMP-9-null
mice (Holmbeck et al. 1999
). The MT1-MMP deficient mice have
craniofacial dysmorphisms caused by impaired intramembraneous bone
formation and dwarfism that is most likely secondary to impaired endochondral ossification. Surprisingly, they also have osteopenia, arthritis, and fibrosis of the soft tissues. Thus, too little MMP can
lead to tissue destruction, a role hypothesized for too much MMP. The
abnormal cranial morphogenesis is partly due to impaired removal of the
cavarial cartilage primordia, which persist and transform into a
fibrotic vestige. This may interfere with the normal formation of the
calvarial bones and suture closure. In the long bones, there is
defective ossification of the epiphysis. In the epiphysis, hypertrophic
cartilage is formed in the center from the inward maturation of
chondrocytes. This hypertrophic cartilage is then ossified by the
invasion into the cartilage of perichondrial vessels through vascular
canals formed from the perichondrium. In the MT1-MMP-deficient mice,
no vascular canal is formed. Therefore, ossification is delayed and
only occurs when the hypertrophic cartilage zone has expanded to
perichondrium, by the invasion of vessels directly from the
perichondrium. Following this delayed ossification of the epiphysis,
the growth plates become disorganized with decreased proliferation of
chondrocytes. This may lead to the dwarfism seen in these mice.
Progressive fibrosis develops in periskeletal soft tissues in these
mice, including the periosteum, tendon, ligament, and capsular
insertions. These may limit bone growth leading to dwarfism. Similar
defects have been observed in mice with a mutation in the collagenase cleavage site of type I collagen (Liu et al. 1995
). Bone marrow stromal
cells and fibroblasts isolated from the MT1-MMP-null mice have
impaired collagenolytic activity in an in vitro assay, suggesting that
the developmental defects in these mice may result from an impairment
in the remodeling of the skeletal matrix as well as of the periskeletal
soft tissues. Of interest is the finding that there is decreased
osteogenesis from MT1-MMP bone marrow stromal cells, suggesting that
the differentiation or function of osteoblasts may also be impaired. In
addition, there is increased osteoclast number in these mice,
indicating that their differentiation and recruitment may also be
affected in MT1-MMP deficiency. This effect of MT1-MMP may not be
secondary to its collagenolytic activity but could conceivably be due
to its activity on growth or differentiation factors for these cells.
This remains to be shown.
| |
Conclusions |
|---|
|
|
|---|
Discovering the roles of the MMPs in physiological processes has not been simple. In vitro studies suggest that MMPs can affect fundamental cellular processes such as proliferation, survival, migration, and morphogenesis. These are the key processes of development. Yet, except for the ones discussed above, developmental defects resulting from inactivation of other MMP genes have not been apparent. In addition, many processes that are affected in vitro by alteration of MMP activity do not seem to be affected by the lack of the relevant MMP in vivo. For example, in vitro kidney development shows a dependency on MMP-9 and MT1-MMP. Yet, kidney development is not greatly affected in either the MMP-9 or MT1-MMP knockout mice. It is possible that there are compensatory mechanisms in vivo that are not available in the in vitro system. Inactivating individual MMPs may show defects in processes in which the particular MMP play an indispensable function, but may not reveal minor functions. In addition, it is usually difficult to decipher the mechanisms of action of the relevant MMP from looking at the end result of its deficiency in what is usually a very complex interplay of many cellular processes.
Another fundamental question in the function of MMPs is that of the relevant in vivo substrates of MMPs. It is clear that fundamentally these enzymes exert their effects by degrading either ECM or other molecules, but what these are in each specific case has eluded discovery. Even though we have gained significant insights into how MMPs may function, more work needs to be done before the roles of MMPs in development are fully elucidated. Inactivation of other MMPs, inactivation of multiple MMPs, and discovering the relevant substrates will be the focus of future studies. As more data are generated, the pieces may eventually come together.
| |
Acknowledgments |
|---|
This work was supported by grants from the National Institutes of Health (CA57621, CA72006, AR46238, DE13058), by a grant form the Human Frontiers Science Program (RG0051/1999M), and by a Mentored Physician-Scientist Award (K08 HL03880).
| |
Footnotes |
|---|
3 Corresponding author.
E-MAIL thiennu{at}itsa.ucsf.edu; FAX (415) 206-4123.
Article and publication are at www.genesdev.org/cgi/doi/10.1101/gad.815400.
| |
References |
|---|
|
|
|---|
by matrix metalloproteinases.
J. Biol. Chem.
271:
14657-14660