| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Commentary |
From the Fred Hutchinson Cancer Research Center, Seattle, Washington
Enzymes typically function either intracellularly or extracellularly, but not at both locations. In this issue of The American Journal of Pathology, Si-Tayeb and colleagues demonstrate that an active form of the well-known extracellular protease matrix metalloproteinase-3 (MMP-3, also known as stromelysin-1) is also present in the nuclear compartment of malignant and nontransformed hepatocytes.1 Although at first this finding might be dismissed as sloppy protein targeting, the identification of a nuclear localization signal in MMP-3 suggests that this alternate localization represents an important, heretofore underappreciated, aspect of MMP-3 function.
Matrix metalloproteinases represent a large family of vertebrate proteins best known for their functions in remodeling of extracellular matrix and for their important roles in wound healing, angiogenesis, and invasive properties of cancer cells.2
Cell membrane proteins such as E-cadherin, protransforming growth factor-ß, and pro-tumor necrosis factor-
have also been identified as MMP substrates, expanding the potential importance of this family to include direct effects on cell-cell signaling and intercellular interactions.3-5
MMPs are part of the larger metzincin superfamily of proteins distinguished by a catalytic zinc-binding motif with three histidines and an undergirding methionine.6
The MMPs are produced as inactive zymogens with a prodomain containing a "cysteine switch" that coordinates the active site zinc.7
Extracellular processing occurs in sequential steps, with an initial cleavage by a serine protease (or another activated MMP) within the prodomain, followed by a second intermolecular cleavage by an active MMP to generate the mature enzyme.8
While investigating MMP-3 expression in hepatocellular carcinoma, Si-Tayeb et al observed prominent homogeneous nuclear immunostaining for MMP-3 in 17:19 hepatocellular carcinoma specimens.1 Adjacent hepatocytes and myofibroblasts also exhibited nuclear MMP-3 staining, although plasmocytes featured the expected cytoplasmic staining. This finding was confirmed with two MMP-3 antibodies directed at the hinge and proximal hemopexin domains but not with antibodies recognizing N- or C-terminal regions of MMP-3. Thus, the nuclear form(s) of MMP-3 appeared to have undergone additional processing or to have concealed certain epitopes. Western blot analysis of the HepG2 hepatocellular carcinoma cell line revealed 45- and 35-kd immunoreactive MMP-3 proteins in nuclear fractions, consistent with mature (45-kd) and truncated versions of the enzyme. Surprisingly, casein zymography and casein affinity purification studies in HepG2 nuclear extracts revealed only the 35-kd form. The authors suggest that the 45-kd MMP-3 may be maintained in an inactive state by forming a complex with another protein. Notably, TIMP1 (tissue inhibitor of metalloproteinase-1) has been observed in cell nuclei.9,10
There are previous examples of intracellular activation for certain MMPs due to the presence of a furin-cleavage site at the junction of the propeptide and mature enzyme in MMP-11, MMP-27, and the four membrane-type-MMPs.11 Furin, a subtilisin family endopeptidase localized to the trans-Golgi compartment, possesses strong preference for basic amino acids at positions P4 to P1 of the cleavage site (RXXR). MMP-3 and MMP-2 have furin cleavage motifs immediately preceding the cysteine switch residue in their prodomains. Surprisingly, furin-dependent intracellular cleavage of MMP-2 was recently reported to generate an inactive enzyme in Cos cells, suggesting that additional processing was required for MMP activation.12 The most likely candidates for secondary intracellular processing of MMP-2 and/or MMP-3 are the membrane-type-MMPs or other MMPs capable of direct activation by furin. An enhanced green fluorescent protein-tagged mature MMP-3 protein with an active site mutation retained nuclear localization in the Si-Tayeb study, suggesting that self-processing is unlikely to play a role in MMP-3 activation.1
Si-Tayeb and colleagues identified a putative nuclear localization signal in the catalytic domain of MMP-3.1 The addition of this domain to an enhanced green fluorescent protein reporter increased nuclear localization, whereas mutation or partial deletion of this sequence in the enhanced green fluorescent protein-MMP-3 fusion protein interfered with nuclear localization. The absence of full-length MMP-3 in nuclear extracts raises the possibility that processing is required to expose the nuclear localization signal for nuclear transport. Golubkov et al have described trafficking of MT1-MMP to centrosomes after cell surface expression and endocytic trafficking, and this possibility has not been excluded for MMP-3.13
Transfection of HepG2 cells with enhanced green fluorescent protein-MMP-3 elicited an apoptotic response characterized by activated caspase-3.1
The lack of apoptosis with the previously mentioned MMP-3 active site mutant and suppression of apoptosis with an MMP-3 inhibitor suggest that specific proteolytic targets are cleaved by MMP-3. Recent articles have demonstrated the presence of active MMP-2 in the nuclei and sarcomeres of cardiac myocytes and identified nuclear poly(ADP-ribose) polymerase as a novel MMP-2 substrate.14,15
Known substrates for MMP-3 include MMPs (MMP-1, -3, and -9), insulin and insulin-like growth factor binding proteins, substance P, protease inhibitors (2-macroglobulin, antithrombin III, 1-antichymotrypsin, 1-proteinase inhibitor), urokinase-type plasminogen activator, and extracellular matrix components (nonfibrillar and denatured fibrillar collagens, aggrecan, fibronectin, laminin, and elastin).16
MMPs recognize at least six amino acids with different degrees of conservation at the substrate cleavage site, complicating bioinformatics efforts to identify substrates.16,17
Nuclear targets for proteolysis during apoptosis include poly(ADP-ribose) polymerase, lamins, inhibitor of caspase-activated DNase, and various factors involved in RNA processing and DNA repair, none of which are known to function upstream of caspase-3 in apoptosis models. In contrast, cleavage of Rb18,19
and several kinases with nuclear localization, including mammalian Ste20-like kinase 1-3 (Mst-1/2,3)20-22
and p21-activated protein kinase-2,23
are sufficient to promote apoptosis. Another possibility is cleavage of an initiator caspase, several of which have been observed in the nucleus (caspase-2, -8, -9, and -10).24-26
Since processed, active MMP-3 is identified in nuclei of apparently healthy cells, MMP-3 activity may be tolerated in nonapoptotic roles.
One of the most interesting aspects of MMP activity in cancer models is the association between MMP activation and genetic instability. Several transgenic mouse models have demonstrated the ability of individual MMPs to promote early stages of carcinogenesis rather than the expected late effects on metastatic/invasive phenotypes.3,27,28 The p53 and c-abl responses to DNA damage are positively modulated by cell adhesion.29,30 Interference with this mechanism by MMP cleavage of extracellular matrix proteins may result in a failure to repair damaged DNA or eliminate cells with irreparable damage through apoptosis. A second potential mechanism for MMP-dependent genetic instability, involving oxidative stress, was identified by Radisky et al.31 They identified an isoform of the Rac1b small G protein that is induced in cells expressing active MMP-3 and is both sufficient and necessary for mitochondrial ROS generation.
Last, the paper by Si-Tayeb et al provides a third potential mechanism of MMP-3-associated genetic instability. The possibility that apoptotic DNA fragmentation itself is a cause of genetic instability is doubtful, since cells deficient in the apoptotic endonuclease caspase-activated DNase have higher, not lower, levels of mutations, gene amplifications, and chromosomal aberrations.32 However, cleavage of nuclear proteins, possibly those involved in DNA repair, might lead to accumulation of DNA base lesions or strand breaks, eventually triggering apoptosis. Combined with MT1-MMP-induced centrosome dysfunction,13 regulation of genomic integrity by nuclear substrates of MMP-3 would expand the list of potential tumorigenic targets of MMPs. In the face of these multiple pathways, which events are most relevant to cancer models? The identification by Si-Tayeb et al of a nuclear localization signal domain in MMP-3 provides a logical means to begin dissecting such relevant downstream targets.
Footnotes
Address reprint requests to David M. Hockenbery, Fred Hutchinson Cancer Research Center, Division of Clinical Research and Human Biology, 1100 Fairview Avenue North, C3-168, Seattle, WA 98109-1024. E-mail: dhockenb{at}fhcrc.org
This commentary relates to Si-Tayeb et al, Am J Pathol 2006, 169:1390–1401, published in this issue.
Accepted for publication June 8, 2006.
References
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
