Detection of a latent soluble form of membrane type 1 matrix metalloprotease bound with tissue inhibitor of matrix metalloproteinases-2 in periprosthetic tissues and fluids from loose arthroplasty endoprostheses
Laboratory of Biochemistry, Department of Chemistry, University of Patras, Greece
Membrane type 1 matrix metalloproteinase (MT1-MMP) is implicated in pericellular proteolysis, and, together with tissue inhibitor of matrix metalloproteinases-2 (TIMP-2), in the activation of pro-matrix metalloproteinase-2 on the cell surface. It is expressed on the cell surface either activated or as a proenzyme. A soluble form of MT1-MMP (sMT1-MMP) has been previously identified in periprosthetic tissues and fluid of patients with loose arthroplasty endoprostheses. The aim of this study was to examine periprosthetic tissues and fluids from patients with loose arthroplasty endoprostheses, as well as tissues and fluids from patients with other disorders, for the presence of sMT1-MMP, and to investigate its activation state and possible role. With antibody against MT1-MMP, a protein with molecular mass of ~ 57 kDa was detected by western blotting in all samples tested, representing a soluble form of MT1-MMP, which cannot be ascribed to alternative splicing, as northern blotting showed only one transcript. With various biochemical methods, it was shown that this species occurs in a latent form bearing the N-terminal prodomain, and, additionally, it is bound to TIMP-2, which appeared to be bound via its C-terminal domain to a site different from the active site. Cell ELISA and immunohistochemical analysis revealed that, besides fibroblasts, all other cells, such as inflammatory, epithelial, endothelial, giant and cancer cells, express MT1-MMP on their plasma membrane as a proenzyme. Taking into account the proteolytic abilities of MT1-MMP, the latent sMT1-MMP–TIMP-2 complex could be considered as a new interstitial collagenase. However, the exact role, the production mechanism and the cell origin of this complex remain to be elucidated.
Total hip or knee replacement is an operation whereby the diseased cartilage and bone of the hip or knee joint is surgically replaced with artificial materials. Aseptic loosening of endoprostheses represents the predominant complication of this operation, which usually occurs 7–10 years after the primary arthroplasty. Despite intensive research, the exact biological mechanism responsible has not yet been completely elucidated. Today, it is known that particles from fragmentation and wear of the implanted materials, released into the periprosthetic tissues and fluids, cause foreign body reactions, and activate cells, mainly macrophages, to produce a variety of cytokines, growth factors, prostaglandins, and proteolytic enzymes. These cell products, through complex mechanisms, can induce periprosthetic osteolysis that exceeds the reparative capacity of the fibrous and osseous tissues, resulting in loosening of endoprostheses [1, 2]. The proteolytic enzymes, and in particular the extracellular matrix metalloproteinases (MMPs), appear to play a very important role in this process. The presence of MMPs, as well as of their tissue inhibitors, tissue inhibitor of matrix metalloproteinases-1 (TIMP-1) and tissue inhibitor of matrix metalloproteinases-2 (TIMP-2), in periprosthetic tissues and fluids from loose arthroplasty endoprostheses, has been previously reported [3-5].
MMPs constitute a family of zinc endopeptidases that degrade almost all of the extracellular matrix (ECM) components, process growth factors, growth factor-binding proteins, and cell surface proteins, and play pivotal roles in many normal and pathological processes . Membrane-type MMPs (MT-MMPs), the more recent subgroup of MMPs, appear to play multiple roles. They activate other MMPs, directly cleave some ECM macromolecules, control cell migration, cell signaling, and cell proliferation, and regulate the multiple stages of tumor progression, including growth and angiogenesis [6-11].
Among MT-MMPs, membrane type 1 MMP (MT1-MMP; MMP-14) is the most abundant and active MT-MMP playing a major role in ECM remodeling, directly by degrading several of its components, and indirectly by activating other MMP proenzymes. MT1-MMP is produced intracellularly as an inactive ~ 63-kDa zymogen that is activated by furin-like convertases . Active MT1-MMP (~ 60 kDa) is then inserted into the plasma membrane with the catalytic domain facing the extracellular space, where it can activate other MMP proenzymes, including proMMP-2, proMMP-9, and proMMP-13, in a process that is both MT1-MMP-dependent and TIMP-2-dependent [12-17], and degrade several ECM components, including collagen types I, II, and III, fibronectin, laminin, vitronectin, fibrin, and proteoglycans, as well as a variety of membrane-anchored adhesion molecules, such as integrins and hyaluronan receptor CD44 [18, 19]. In addition to the intracellular proteolytic activation of proMT1-MMP by furin, several lines of evidence suggest that MT1-MMP is also transported to the plasma membrane in a 63-kDa latent precursor form bearing the propeptide domain [20-22], or in a partially processed 60-kDa latent form . It has been reported that this proMT1-MMP on the cell surface can be activated by plasmin  or urokinase .
It has also been reported that mature MT1-MMP can be shed from the cell surface via a non-autocatalytic process resulting in the release of various active soluble forms that may represent the entire ectodomain, including the catalytic domain [22, 25-29]. MT1-MMP shedding has been described in cultures of human mesangial , breast carcinoma [22, 25-27], lung fibroblast , endothelial [28, 29] and fibrosarcoma [25, 26] cells. Soluble MT1-MMP (sMT1-MMP) forms have also been detected in human sputum and bronchoalveolar lavage fluid , suggesting that shedding occurs in vivo and may represent a physiologically relevant process. However, the precise nature and function of this species is unknown.
The detection of sMT1-MMP (~ 56 kDa) in synovial fluid (SF) and pseudosynovial fluid (PSSF) of patients with rheumatoid arthritis (RA), osteoarthritis (OA) and loose arthroplasty endoprostheses has been previously reported . A protein band of ~ 57 kDa was also previously detected in periprosthetic tissue extracts and PSSF from loose arthroplasty endoprostheses, and was ascribed to sΜΤ1-ΜΜΡ .
In the present study, we aimed to identify the ~ 57-kDa sMT1-MMP in periprosthetic tissues and fluids, to investigate its origin and activation state, and to identify equivalent species in tissues and fluid from patients with various disorders, including nasal polyposis (NP) and lung diseases.
Expression of MT1-MMP in periprosthetic tissues
In order to ascertain the expression of MT1-MMP in periprosthetic tissues from loose arthroplasty endoprostheses, total RNA was isolated from interface tissue around loose femoral components (IFT) and pseudocapsular tissue (PCT) obtained from different patients, and used in RT-PCR analysis with specific primers for MT1-MMP, the amplification products of which were subjected to agarose gel electrophoresis (Fig. 1). In all samples, a 670-bp band was visualized, corresponding to human MT1-MMP, indicating that this metalloproteinase is expressed by cells of periprosthetic tissues.
Detection of a soluble form of MT1-MMP in various tissue extracts and fluids
Samples of ultracentrifugation supernatants of IFT and PCT extracts and PSSF from different patients were subjected to western blot analysis, with polyclonal antibody against the extracellular hinge region of human MT1-MMP (Fig. 2A,B). For comparison, samples of SF from RA patients (SF-RA) and and SF from OA patients (SF-OA), identically treated, were also subjected to western blotting (Fig. 2C,D). As shown in Fig. 2, only one major immunoreactive band, of ~ 57 kDa, corresponding to sMT1-MMP was detected in all periprosthetic tissue extracts and fluids. In order to determine whether sMT1-MMP occurs in other tissues, extracts of human nasal polyp tissues and lung tissues from patients with various lung diseases were subjected to western blot analysis with the same polyclonal antibody as above (Fig. S1A,B). As with previous extracts and fluids, a major immunoreactive band, of ~ 57 kDa, was also detected in all polyp and lung tissue extracts, indicating that sMT1-MMP may represent a common type of MT1-MMP, independently of biological tissue and fluid origin. The observed immunoreactive bands of lower molecular mass (Fig. S1B) may represent degradation products of sMT1-MMP, as previously observed [20, 21, 23]. The double band observed in some samples may be attributable to different glycosylation of sMT1-MMP.
This 57-kDa sMT1-MMP could not be ascribed to alternative splicing, as only one transcript (3.7 kb) was detected by northern blotting with total RNA from IFT (Fig. S2).
Investigation of the activation state of sMT1-MMP
For the purpose of investigating whether or not sMT1-MMP is present in an activated form, representative samples of IFT, PCT, polyp and lung extracts, and PSSF, SF-RA, and SF-OA, were incubated in the presence of excess α2-macroglobulin (100 μg·mL−1) for 1 h at 37 °C, and then subjected to western blot analysis with polyclonal antibody against the extracellular hinge region of human MT1-MMP (Fig. 3). It is known that α2-macroglobulin forms complexes of high molecular mass with proteolytically active but not with latent or inactive proteases, which are resistant to SDS treatment and do not enter polyacrylamide gels. As shown in Fig. 3, in all samples, the intensity of the sMT1-MMP band before and after treatment with α2-macroglobulin remained the same, indicating that sMT1-MMP was not able to form complexes with α2-macroglobulin, suggesting that it occurs either in a latent inactive form or in an activated form in complex with TIMP-2.
It is known that TIMP-2 can bind via its N-terminal inhibitory domain to the active site of activated MT1-MMP on the cell surface, and via its C-terminal region to the hemopexin-like domain of proMMP-2, leading to proenzyme activation [14-16]. Therefore, a sample of IFT extract was subjected to western blot analysis with polyclonal antibody against the extracellular hinge region or against the N-terminal propeptide domain (prodomain) of human MT1-MMP. A recombinant fragment of human MT1-MMP, consisting of the prodomain, catalytic domain, and hemopexin domain, representing the extracellular moiety of MT1-MMP (recombinant sMT1-MMP), was subjected to the same treatment and studied in parallel (Fig. 4). As shown in Fig. 4Α,Β, only one band of ~ 57 kDa, reacting with both antibodies and migrating as the human recombinant sMT1-MMP, was detected in IFT extract, suggesting that sMT1-MMP bears the prodomain. Control experiments to determine the specificity of the antibody against the prodomain of MT1-MMP showed that the antibody used does not cross-react with the prodomains of other MMPs of the same molecular mass, such as MMP-1 and MMP-3 (Fig. 4B). The observed immunoreaction with both antibodies was specific because, when only the second antibody was used, no reaction was detected (not shown). Reaction with antibody against the prodomain of MT1-MMP was also observed when samples of PSSF, SF-RA, SF-OA and nasal polyp and lung tissue extracts were subjected to western blotting. A major immunoreactive band of ~ 57 kDa was detected in all cases (Fig. S3). Note that, in nasal polyp extracts, the immunoreactive bands of lower molecular mass, which were previously considered to be degradation products of the 57-kDa protein band, also bear the prodomain (Fig. S3B), which is in contrast to previous data in the literature showing sequential MT1 proteolysis of the propeptide, and then the catalytic domain, leaving the hemopexin domain. It is possible that proteolytic enzymes present in nasal polyp tissue degrade sMT1-MMP in a different way. Further studies are needed in order to elucidate this further.
However, upon immunoprecipitation of sMT1-MMP from tissue extracts and fluids with polyclonal antibody against the hinge region of MT1-MMP, and western blot analysis with mAb against human TIMP-2, an immunoreactive band of ~ 21 kDa, migrating as human recombinant TIMP-2, was observed in all samples (Fig. 5A). As the polyclonal antibody against the hinge region of MT1-MMP used for immunoprecipitation do not cross-react with TIMP-2 (Fig. 5B), and it did not show nonspecific reactions with mAb against TIMP-2 (Fig. 5A), these results suggest that TIMP-2 coprecipitated with sMT1-MMP because it was already in complex with it. Consequently, it appears that the sMT1-MMP in tissue extracts and fluids exists in a latent form, bound to TIMP-2.
These results were further confirmed by reverse immunoprecipitation with mAbs against TIMP-2 and western blotting with polyclonal antibody against the hinge region of MT1-MMP. Only one immunoreactive band, of ~ 57 kDa, corresponding to sMT1-MMP, was observed in all samples (Fig. 5C).
In order to examine how TIMP-2 is bound to sMT1-MMP, in representative samples of IFT and PCT extracts, and PSSF, proMMP-2 was added in excess, and, after preincubation for 1 h, they were subjected to affinity chromatography on gelatin–Sepharose, as described in Experimental procedures. The wash and elution fractions were analyzed by gelatin zymography (Fig. 6A) and western blotting, with polyclonal antibody against the extracellular hinge region of human MT1-MMP (Fig. 6B). Although all of the MMP-2 gelatinolytic activity was obtained in the gelatin-bound pool (Fig. 6A), sMT1-MMP in all three samples was eluted with equilibration buffer (Fig. 6B), indicating that TIMP-2 was bound to sMT1-MMP via its C-terminal domain, and consequently was not able to interact with proMMP-2. The same was observed with SF-RA, SF-OA, and polyp and lung extracts (Fig. 6C,D). The proMMP-2 used was able to form stable complexes with TIMP-2, which, upon affinity chromatography on gelatin–Sepharose, survived and were obtained in the gelatin-bound pool, as determined by western blotting with mAb against TIMP-2 (Fig. 6E,F). These findings suggest that TIMP-2 may not bind to the active site of MT1-MMP, as it does in the MT1-MMP–TIMP-2–proMMP-2 complexes.
When representative samples of IFT and PCT extracts and PSSF were treated with aminophenyl mercuric acetate (APMA) and subjected to western blot analysis with polyclonal antibody against the prodomain of MT1-MMP, only one immunoreactive band, of ~ 57 kDa, was observed, whose electrophoretic mobility was not altered after pretreatment with APMA (Fig. 7A), indicating that, even in the presence of APMA, the immunoreactive band of ~ 57 kDa retained the propeptide. In contrast, when recombinant sMT1-MMP was treated with APMA and subjected to western blot analysis with polyclonal antibody against the extracellular hinge region of human MT1-MMP, an immunoreactive band of ~ 50 kDa appeared (Fig. 7B), which disappeared after preincubation with α2-macroglobulin (Fig. 7B). With antibody against the prodomain of MT1-MMP, no immunoreactive bands appeared in any case (Fig. 7C). These results suggest that, after destabilization of the prodomain of sMT1-MMP with APMA, TIMP-2 can occupy the active site of MMP through its free N-terminal domain, blocking further autocleavage and removal of propeptide. A similar suggestion has previously been made for the proMMP-2–TIMP-2 complexes .
Different forms of MT1-MMP are expressed in various cell types
As mentioned previously, MT1-MMP is expressed on the cell surface in either an activated form (furin-mediated, lacking the prodomain) or in a latent proenzyme form. It has been reported that in a variety of cells, including chondrocytes, synovial fibroblasts, and some cancer cell lines, it is expressed in an active form [32-40], whereas in another group of cells, including lung fibroblasts, human mesangial cells, and some tumor cell lines, it is expressed as a proenzyme [20-22]. It has also been reported that the expression of MT1-MMP in fibroblasts of various origin is induced by tumor necrosis factor-β (TNF-β) and interleukin (IL)-1β [23, 32, 36]. On the basis of these findings, we studied whether the IL-1β-induced and TNF-α-induced MT1-MMP in fibroblasts bears the prodomain. Thus, human nasal polyps and IFT fibroblasts were cultured in the absence or presence of IL-1β or TNF-α, and the expression of MT1-MMP on the cell surface was ascertained by cell ELISA, as previously described . Using antibody against the extracellular hinge region of human MT1-MMP, expression of MT1-MMP was detected on the cell surface of both fibroblasts, which was enhanced in the presence of IL-1β or TNF-α (Fig. 8A,B). No immunoreaction was observed when an antibody against the prodomain was used, indicating that the MT1-MMP on the cell surface is expressed in an activated form lacking the prodomain. These results were further confirmed by western blot analysis. Extracts of IFT fibroblasts, cultured in the absence or presence of IL-1β or TNF-α, were subjected to western blotting (Fig. 8C). When antibody against the extracellular hinge region of human MT1-MMP were used, a major immunoreactive band of ~ 60 kDa was detected, which became more intense in the presence of IL-1β or TNF-α (Fig. 8C), but when antibody against the prodomain were used, no immunoreaction was observed (not shown). In contrast, when the breast cancer cell line MDA-MB-231, which is known to express MT1-MMP [22, 42], was cultured until confluent and tested for the expression of MT1-MMP by cell ELISA with polyclonal antibody against the hinge region or the prodomain of MT1-MMP, significant immunoreaction was observed with both antibodies (Fig. 8D), indicating that, in these cancer cells, MT1-MMP is expressed on their membranes in the latent form bearing the prodomain, in accordance with previous reported results .
If it is supposed that sMT1-MMP is a proteolytic product of the membrane-bound form, then the soluble form detected in the samples tested may have resulted from the proteolytic processing of MT1-MMP on the membranes of different cell types – besides fibroblasts – that express MT1-MMP on the cell membrane in a latent form.
In order to identify the types of cell expressing MT1-MMP on their membranes as a proenzyme, immunohistochemistry was applied. Paraffin sections were prepared from various tissues, such as synovial membrane (SM), IFT, nasal polyps, and lung cancer, which were immunostained with polyclonal antibody against the extracellular hinge region or the prodomain of human MT1-MMP.
In all cases, both the inflammatory stromal cells and the cancer cells were stained with both antibodies, suggesting that MT1-MMP in these cells is expressed in a latent form bearing the prodomain (Fig. 9). Endothelial and epithelial cells were strongly stained with the antibody against the prodomain (Fig. 9D–G), and slightly or not at all with the antibody against the hinge region domain (Fig. 9A–C). This may be attributable to steric hindrance resulting from the dense arrangement of these cells in endothelium and epithelium. For the same reason, the giant cells, clusters of activated macrophages, were intensely stained with the antibody against the prodomain (Fig. 9D–G) rather than with the antibody against the hinge region (Fig. 9A–C).
In contrast, stromal fibroblasts were stained with the antibody against the hinge region (Fig. 9A–C) and not at all with the antibody against the prodomain (Fig. 9D–G), suggesting that these cells express MT1-MMP in an activated form without the prodomain.
In conclusion, it seems that, besides fibroblasts, all other cells, such as inflammatory cells, endothelial and epithelial cells, and giant cells, express MT1-MMP on their membrane in a latent form, and the soluble form detected in our samples may be the result of the proteolytic processing of this form of MT1-MMP.
In the present study, it has been shown that a soluble form of MT1-MMP (sMT1-MMP) of molecular mass ~ 57 kDa is present in periprosthetic tissue extracts and fluids, SF from patients with OA and RA, and human nasal polyp tissue extracts and lung tissue extracts from patients with various lung diseases. The detection of a type of MT1-MMP (~ 56 kDa) in SF and PSSF, and periprosthetic tissue extracts from patients with RA, OA, and loose arthroplasty endoprostheses, has been previously reported, and was ascribed to a soluble form of ΜΤ1-ΜΜΡ [4, 5]. Soluble forms of MT1-MMP were also detected in human sputum and bronchoalveolar lavage fluid , but until now there have been no available data on the presence of a respective form of MT1-MMP in human nasal polyp tissues. This detected form of ΜΤ1-ΜΜΡ represents, in fact, a soluble type of MT1-MMP, as all tissues used were extracted in the absence of a detergent, and the extracts, as well as the various biological fluids used, were subsequently centrifuged at 100 000 g. Thus, cells and cell membrane particles containing the trans-membrane MT1-MMP were removed in the precipitates, and therefore only a soluble form of MT1-MMP would be expected to exist in the 100 000 g supernatants.
By the application of various biochemical methods, it was shown that the sMT1-MMP detected in all samples is in the form of proenzyme, with exactly the same molecular mass as a recombinant fragment of human MT1-MMP that consists of the propeptide, catalytic and hemopexin domains, representing the extracellular moiety of MT1-MMP. In addition, it was shown that TIMP-2 is bound to this sMT1-MMP proenzyme via its C-terminal domain to a site other than the active site of the proenzyme.
Northern blot analysis on IFT showed only one mRNA of 3.7 kb, of the same size as the corresponding mRNA reported in osteoblasts (~ 3.5 kb) . One mRNA for MT1-MMP of 4.5 kb was also detected previously in various tissues [21, 33, 44, 45]. Thus, the sMT1-MMP detected in all samples cannot be ascribed to alternative splicing, and is probably derived from a respective membrane-associated form. Cell ELISA and immunohistochemical analysis of SMs, IFT, nasal polyp tissue and lung cancer tissue, with specific antibody against the extracellular hinge region or the prodomain of MT1-MMP, revealed that, besides fibroblasts, all other cells, such as inflammatory, epithelial, endothelial, cancer and giant cells, express MT1-MMP on the cell surface in the form of a proenzyme. These observations indicate that the soluble form detected in our tissue extracts and fluids possibly originates from this membrane-associated latent form of MT1-MMP in these cells. However, we are not in a position to know, whether the MT1-MMP proenzyme on the cell surface carries TIMP-2 or not.
It must be noted that, although fibroblasts express MT1-MMP in an activated form, a respective soluble form of MT1-MMP was not detected in all samples tested. All of the detected soluble forms with a molecular mass of < 57 kDa contained the prodomain. It seems that the soluble activated forms of MT1-MMP are unstable, and are further degraded in the extracellular space.
Various forms of sMT1-MMP, containing or not containing the prodomain, have been detected in several cell culture-conditioned media and biological fluids [21, 22, 25-30]. A latent form of sMT1-MMP that exists in complex with TIMP-2 is reported here for the first time. Imai et al. have described secretion into the culture medium of a human breast carcinoma cell line of a 56-kDa propeptide-deleted MT1-MMP complexed to TIMP-2 . The existence of a soluble form of MT1-MMP was also suspected by Okada et al.  who found, in human carcinomas, that MT1-MMP transcripts were present in stroma cells but not in cancer cells, whereas the protein was detected in cancer cells. Therefore, they suggested that MT1-MMP may be an ectoenzyme generated by cleavage on the cell surface.
It has been reported MT1-MMP is synthesized as an inactive zymogen with a molecular mass of 63 kDa and is activated intracellularly by furin, resulting in the activated form of ~ 57 kDa. Examination of the profile of MT1-MMP forms in cell extracts showed that, in addition to the 57-kDa active species, MT1-MMP is also found as a series of forms ranging from ~ 44 kDa to 40 kDa [15, 48]. Both the 57-kDa and 44-kDa MT1-MMP forms have been detected in many cell types expressing the natural enzyme, but they are also found after expression of recombinant MT1-MMP.
The presence of the 44-kDa product was associated with enhanced MT1-MMP activity. In addition to furin-mediated intracellular activation, a large number of studies have shown that MT1-MMP is also transferred to the cell membrane as a proenzyme. MT1-MMP in transfected cells, human fibroblasts and human HT-1080 fibrosarcoma cells is expressed as an inactive proenzyme with a molecular mass of 63 kDa, or as a partially hydrolyzed inactive form of 60 kDa [28, 49, 50]. The cell membrane-associated latent form of MT1-MMP can be activated by plasmin or urokinase-type plasminogen activator (uPA) [21, 24]. However, other investigators have reported that the removal of propeptide from the latent membrane-associated MT1-MMP is not required for its function concerning the activation of proMMP-2. Indeed, Cao et al. suggested that the propeptide domain of MT1-MMP is required for binding to TIMP-2 and for the activation of proMMP-2 . Their experimental data were consistent with their hypothesis that conformational effects induced by the plasma membrane provide functional activity to membrane-bound MT1-MMP without cleavage of the molecule. In contrast, using detergent extracts of crude plasma membranes, they confirmed that the prodomain of MT1-MMP is not required for function of sMT1-MMP . Li et al.  showed, in breast cancer cell lines, the presence of membrane-bound MT1-MMP bearing the prodomain, as well as the presence of a soluble form of MT1-MMP with a molecular mass of 57 kDa, also bearing the prodomain, in the culture medium. These two forms, membrane-bound and soluble, according to the results of the study, were able to activate the proenzyme of MMP-2 without removal of the prodomain. They suggested that the microenvironment of the membrane favored the activation of MT1-MMP through changes in its conformation, without actual removal of the prodomain. They also proposed that the soluble form is derived from the membrane-bound form by proteolysis. In other studies, it has been suggested that the prodomain of MT1-MMP can function as an intramolecular chaperone, being responsible for the safe transport of MT1-MMP on the cell surface [52, 53].
Taking the above literature reports and our data together, it may be suggested that MT1-MMP exists in at least 10 forms: five membrane-associated forms (a furin-activated 57-kDa form, a furin-activated 57-kDa form bound to TIMP-2, a furin-activated and autocatalytically degraded to a 44-kDa inactive form, a latent proenzyme of 63 kDa, and a latent proenzyme of 63 kDa bound to TIMP-2); and five respective soluble forms (a 50-kDa activated form, a 50-kDa activated form bound to TIMP-2, a 37-kDa inactive form, a latent proenzyme of 57 kDa, and a latent proenzyme of 57 kDa bound to TIMP-2). As mentioned above, the soluble forms of MT1-MMP cannot be ascribed to alternative splicing. Therefore, all soluble forms of MT1-MMP have been considered to be ectoenzymes derived from the respective membrane-bound forms by proteolysis, although the protease that generates these soluble forms has remained, until now, elusive [22, 46-48]. However, another mechanism of soluble form generation, such as RNA editing, cannot be excluded.
Regarding the sMT1-MMP–TIMP-2 complex in our study, we cannot be definite about the way in which this complex is formed. It is possible that it is formed on the cell surface, as previously described by Cao et al. , and is subsequently released into the extracellular space through proteolysis, or it is formed in the extracellular space after release of the latent soluble form of MT1-MMP from the cell surface, as previously described by Li et al. .
It is known that both MT1-MMP and proMMP-2 are able to activate other MMPs and to process various ECM molecules . Taking into account these data, the various forms of sMT1-MMP could be considered as additional interstitial collagenases.
The latent soluble form of 57 kDa in complex with TIMP-2 detected in our study resembles the complex of proMMP-2 with TIMP-2, because, among MMPs, only the zymogens of gelatinases A and B occur in complexes with TIMP-2 and TIMP-1, respectively. However, as mentioned above, APMA had no effect on the activation of sMT1-MMP. This might be because of the occupation of the active site of the free N-terminal domain of TIMP-2 immediately after destabilization of the propeptide. Subsequently, even if the proenzyme of sMT1-MMP in complex with TIMP-2 can be activated by propeptide removal, in response to a number of activators, such as plasmin and uPA [21, 24], TIMP-2 present in the complex, with its N-terminal domain free, is able to recover the active site, leading to inhibition of the activated proenzyme, as previously observed with the proMMP-2–TIMP-2 and proMMP-9–TIMP-1 complexes [31, 54]. On the other hand, the latent sMT1-MMP–TIMP-2 complex could play an inhibitory role, as the free N-terminal domain of TIMP-2 can interact with the active sites of other activated MMPs, as previously described for the complexes of gelatinases with tissue inhibitor of MMPs [31, 54, 55]. However, the blockade of TIMP-2 of the complex with an activated MMP in turn renders the latent sMT1-MMP activatable by various activators, such as plasmin, uPA, and other MMPs, suggesting that the latent sMT1-MMP–TIMP-2 complex may regulate ECM breakdown in tissues by switching the predominant MMP activity from one type to another. The same has previously been reported for the proMMP-2–TIMP-2 and proMMP-9–TIMP-1 complexes [31, 54].
The conclusions concerning the exact role, the production mechanism and the cell origin of the latent sMT1-MMP–TIMP-2 complex identified in this study need to be confirmed by further research, with a larger cohort of patients, and possibly samples of different biological origin. These are under consideration.
All tissue specimens and fluids were collected in the Departments of Orthopedics and Otorhinolaryngology, University Hospital of Patras, Greece. The methods used conformed to the standards set by the Declaration of Helsinki. The study was approved by the local Ethics Committee on human experimentation of the University Hospital of Patras, and informed consent was obtained from each patient before the surgical procedure.
Protein A–Sepharose 4B, protein G–agarose, gelatin–Sepharose 4B, gelatin, α2-macroglobulin and peroxidase-conjugated goat anti-(mouse IgG) were purchased from Sigma Chemical Co. (St Louis, MO, USA); peoxidase-conjugated goat anti-rabbit IgG, polyclonal antibodies against human MT1-MMP (hinge region and prodomain), mAbs against TIMP-2 and human recombinant sMT1-MMP, proMMP-1, proMMP-2, proMMP-3 and TIMP-2 were purchased from Chemicon (Temecula, CA, USA). The mAb against TIMP-2 (T2-101) used in immunoprecipitation experiments, which was produced against an epitope that is not involved in binding of TIMP-2 to proMMP-2, was purchased from Abcam (Cambridge, UK). All other reagents were of analytical grade.
Patients and samples
Samples of IFT, PCT and PSSF were obtained from patients undergoing revision for aseptic total hip or knee replacement loosening. IFT is a synovial membrane-like granulomatous tissue that develops at the bone-cement–prosthesis interface. Immunohistochemical studies of this tissue have revealed an abundance of macrophages, fibroblasts, giant cells, neutrophils, and lymphocytes. PCT is the tissue around the pseudojoint cavity formed around the joint implants, and PSSF is the fluid filling this pseudojoint cavity. PCT may play a role as a producer of proteolytic enzymes and other factors by releasing them into PSSF, where they have access to IFT. Subsequently, these proinflammatory factors, gelatinases and proteases contribute to periprosthetic osteolysis and to failure of the joint implant. SMs and SFs were collected from patients with OA or RA undergoing total knee replacement. Nasal polyp tissues were obtained from patients suffering from chronic sinusitis with polyposis, subjected to sinus surgery. NP is a chronic inflammatory disease of the upper airways, featuring inflammatory cell infiltration, modifications of epithelial differentiation, and tissue remodeling, including basement membrane thickening, gland modifications, ECM accumulation, and edema. It has been reported that polyp formation involves ECM swelling (protrusion) through an initial localized epithelial defect. Interactions between epithelial, stromal and inflammatory cells could then result in further polyp growth. Mucosal inflammation, especially containing an infiltration of eosinophils, is probably the most important factor in the development of NP . The polyp stroma contains various mediators, including cytokines, growth factors, adhesion molecules, and proteolytic enzymes. The extracts of lung tissues obtained from patients with various lung diseases, such as cancer, sarcoma, pulmonary fibrosis, connective tissue disease, and tuberculosis, were kindly provided by E. Papakonstantinou (Department of Pharmacology, School of Medicine, Aristotle University of Thessaloniki, Greece).
Immediately after aspiration from the joint, the PSSF, SF-RA and SF-OA samples were centrifuged at 12 000 g for 10 min at 4 °C to remove cells and debris. Samples of tissues were minced into small pieces, and homogenized with a polytron homogenizer in extraction buffer (50 mm Tris/HCl, pH 7.5, 10 mm CaCl2, 2 m KCl; 3 mL buffer per g tissue) containing the proteinase inhibitors phenylmethanesulfonyl fluoride (1 mm), N-ethylmaleimide (10 mm), and benzamidine-HCl (10 mm), in an ice bath. After stirring for 1 h at 4 °C, the homogenates were centrifuged at 10 000 g for 30 min at 4 °C. Both these supernatants and the supernatants from centrifugation at 12 000 g of PSSF, SF-RA and SF-OA samples were adjusted to contain a final concentration of 250 mm sucrose, and were then centrifuged at 100 000 g for 1 h at 4 °C. The supernatants were collected and enriched in MMPs by precipitation with (NH4)2SO4 (60% saturation). The resulting precipitates were dissolved in enzyme buffer (50 mm Tris/HCl, pH 7.5, 0.2 m NaCl, 10 mm CaCl2, 0.02% NaN3), dialyzed against the same buffer, and used for the assays.
Primary fibroblast cultures
Fibroblasts from IFT membranes were isolated by Clostridium collagenase digestion as previously described . The cell cultures were performed according to Pavlaki et al. . For the isolation and culture of the nasal polyp fibroblasts, the tissues, after removal of the epithelial layer, were rinsed and minced, and the fragments were placed in plastic flasks and grown in a humidified 5% CO2 atmosphere in DMEM (BioChrom AG, Berlin, Germany) supplemented with 10% fetal bovine serum, and 1% penicillin and streptomycin. After fibroblasts had migrated from tissue explants and become confluent, cells were trypsinized, split at a 1 : 3 ratio, and recultured in the same medium.
Cells from passages 3–5 were seeded onto six-well plates (Greiner, Frickenhausen, Germany) and cultured until confluent. Then, cells were starved in serum-free DMEM containing 0.2% lactalbumin hydrolysate (Sigma-Aldrich Chemical Co., St Louis, MO, USA)  for 24 h before the experiment, with the experiment being conducted in the same medium in the absence or presence of the cytokine being tested.
The cell ELISAs were performed according to Simiantonaki et al. . Incubation with rabbit polyclonal antibodies against human MT1-MMP-1 (hinge region or prodomain) (4 μg·mL−1) and incubation with the second antibody [peroxidase-conjugated goat anti-(rabbit IgG)] at a dilution of 1 : 1000 were performed at 37 °C for 2 h.
The samples, dissolved in Laemmli sample buffer , were treated with 2-mercaptoethanol and then subjected to SDS/PAGE on 12% polyacrylamide gels; this was followed by western blot analysis as previously described [5, 58]. Incubation with rabbit polyclonal antibodies against human MT1-MMP (hinge region or prodomain) (1 μg·mL−1) or mouse mAb against TIMP-2 (3 μg·mL−1) was performed overnight at 4 °C. Incubation with peroxidase-conjugated goat anti-(rabbit IgG) at a dilution of 1 : 4000 (for MT1-MMP) or peroxidase-conjugated goat anti-(mouse IgG) at a dilution of 1 : 2000 (for TIMP-2) was performed at room temperature for 2 h. In order to avoid false results because of nonspecific reactions of the second antibody with human IgG, which may be present in tissue extracts and fluids, before use this antibody was adsorbed with human IgG in excess. Immunoreactive proteins were detected with the enhanced chemiluminescence method, according to the manufacturer's instructions (Pierce, Rockford, IL, USA).
In representative samples of tissue extracts and fluids, containing ~ 100 μg of total protein, in NaCl/Pi-Tween, 2 μg of polyclonal antibody against the hinge region of ΜΤ1-ΜΜΡ, or 2 μg of mAb against TIMP-2 (T2-101), was added per sample, and this was followed by incubation overnight at 4 °C. Then, each sample was applied to a protein A–Sepharose CL-4B gel (when polyclonal antibody against the hinge region of ΜΤ1-ΜΜΡ were used), or a protein G–agarose gel [when mAb against TIMP-2 (T2-101) was used], packaged in an Eppendorf centrifugation flex-tube of 1.5 mL, equilibrated in NaCl/Pi-Tween, and incubated at room temperature for 1 h, with mild stirring. After extensive washing with NaCl/Pi-Tween, the gel was precipitated by centrifugation at 10000 g for 5 min., and treated with an equal volume of Laemmli sample buffer  in the presence of 2-mercaptoethanol, and the supernatant was subjected to western blotting, as described above, with mAb against TIMP-2 (3 μg·mL−1), or polyclonal antibodies against the hinge region of ΜΤ1-ΜΜΡ (3 μg·mL−1), in order to determine whether sMT1-MMP was in complex with TIMP-2.
Gelatin zymography was performed as previously described [5, 58].
Extraction of RNA from tissues and RT-PCR
Total RNA was extracted from periprosthetic tissues with the RNAeasy PROTECT kit (Qiagen, Stanford, Valencia, USA), according to the manufacturer's instructions (RNeasy Mini Handbook, Qiagen, 2001). Up to 100 mg of tissue was used. RT-PCR was carried out with the One-Step RT-PCR kit from Qiagen (Stanford, Valencia, USA). Total RNA (20 ng) was added to each reverse transcription reaction, containing the kit mix and the appropriate PCR primers (20 pmol) in a total volume of 50 μL. Reverse transcription was carried out for 30 min at 50 °C, and this was followed by a 15-min step at 95 °C. Amplifications were performed with 30 cycles. Each cycle included denaturation at 95 °C for 1 min, annealing at 60 °C for 1 min, and extension at 72 °C for 1 min, followed by a final extension at 72 °C for 10 min. The PCR primers used for human MT1-MMP (sense, 5′-GAT-AAA-CCC-AAA-AAC-CCC-ACC-TA-3′; antisense, 5′-CCC-TCC-TCG-TCC-ACC-TCA-ATG-3′)  were synthesized by MWG-Biotech AG (Ebersberg, Germany). The PCR amplified a 670-bp product, which was analyzed and visualized by electrophoresis on a 1.5% agarose gel in the presence of SYBR Gold nucleic acid gel stain (Molecular Probes, Invitrogen, USA). The PCR products of 10 reactions were combined, radioactively labeled with 32P, and used as a probe for northern blot analysis after sequencing.
Northern blot analysis
Total RNA was isolated from IFT, as described above. Thirty micrograms of total RNA was denatured with formamide, separated by electrophoresis on a 1% agarose gel, and blotted onto a nylon membrane (Zeta-Probe; Biorad). Prehybridization was performed for 30 min at 43 °C in a solution containing 50% formamide, 0.12 m sodium phosphate (pH 7.2), 0.25 m sodium chloride, 7% SDS, and 1 mm EDTA. The membrane was hybridized with a 32P-labeled 670-bp human MT1-MMP cDNA obtained by RT-PCR (see above) at 43 °C for 16 h, and washed twice in 2 × standard sodium citrate and 0.1% SDS buffer, under high-stringency conditions, for 30 min at room temperature. The film was developed after 72 h of exposure.
Immunohistochemistry was performed as previously described . Incubation with polyclonal antibodies against human MT1-MMP (hinge region or prodomain) (10 μg·mL−1) was performed overnight at 4 °C. Binding was detected with the avidin–biotin–peroxidase complex method .
Affinity chromatography on gelatin–Sepharose
Gelatin–Sepharose affinity chromatography was performed as previously described .
Treatment with APMA
Treatment with APMA was performed as previously described [5, 60].
Treatment with α2-macroglobulin
Samples of tissue extracts and fluids, previously subjected to various treatments, were then incubated in enzyme buffer (50 mm Tris/HCl, pH 7.5, 0.2 m NaCl, 10 mm CaCl2, 0.02% NaN3) in the absence or presence of α2-macroglobulin in excess (100 μg·mL−1) at 37 °C for 1 h, and then analyzed by western blotting.
The protein content of samples was determined by the method of Bradford, with BSA as standard .