2.3. The lysyl oxidase isoenzyme LOX
2.3.1. Molecular properties
Lysyl oxidase enzyme (LOX, protein-lysine 6-oxidase, EC 1.4.3.13) was initially believed to be of one type only. Studies on this enzyme were seriously limited for a long time by its marked insolubility and aggregation tendency. A major breakthrough for studies on lysyl oxidase was the discovery that the enzyme can be subsequently solubilized by using buffers containing 4-6 M urea, and its activity can be recovered by the removal of the urea (Narayanan et al. 1974). Lysyl oxidase was then purified from several different tissues and organisms, including chick cartilage (Stassen 1976), bovine aorta (Kagan et al. 1979, Cronlund & Kagan 1986) and lung (Cronlund & Kagan 1986), human placenta (Kuivaniemi et al. 1984), and piglet skin (Shackleton & Hulmes 1990). The molecular mass of the enzyme from all those sources was found to be approximately 32 kDa. Interestingly, lysyl oxidase extracted from several tissues resolved upon DEAE chromatography into multiple forms (Kagan 1986). Furthermore, at least four forms obtained from chick cartilage (Stassen 1976) and human placenta (Kuivaniemi et al. 1984) had similar substrate specifities, amino acid compositions, and molecular masses of approximately 32 kDa (Kagan 1986). These findings suggested the existence of alternatively spliced or other forms of a single lysyl oxidase, or the presence of isoenzymes with similar but not identical properties.
The lysyl oxidase isoenzyme LOX has now been cloned from rat (Trackman et al. 1990, 1991), human (Hämäläinen et al. 1991, 1993, Mariani et al. 1992), chick (Wu et al. 1992), and mouse (Contente et al. 1993) tissues. The mRNA for human LOX is found as multiple species with sizes of 5.5, 4.3, 2.4, and 2.0 kb due to the use of alternate polyadenylation sites, and partially to the existence of multiple transcription initiation sites (Hämäläinen et al. 1991, Mariani et al. 1992, Boyd et al 1995). The biological relevance of these transcripts, however, is unknown. The human LOX gene is located on chromosome 5q23.3-31.2 (Hämäläinen et al. 1991, 1993, Mariani et al. 1992, Boyd et al. 1995) and encodes a 417 amino acid polypeptide, of which the first 21 residues correspond to the signal peptide (Hämäläinen et al. 1991, Mariani et al. 1992). The mouse Lox gene has been mapped to chromosome 18 (Mock et al. 1992, Chang et al. 1993, Contente et al. 1993).
2.3.1.1. Biosynthesis and processing of the LOX precursor
The LOX protein is known to be secreted into the extracellular space (Layman et al. 1972, Byers et al. 1980, Royce et al. 1980, Peltonen et al. 1983, Kuivaniemi et al. 1986a). The transport through the Golgi and to the cell membrane probably takes place in vesicles formed by budding of, and fusion with, the subcellular membranes (Kosonen et al. 1997, Rucker et al. 1998). The first evidence for the biosynthesis of a LOX proenzyme was obtained by immunoprecipitation of a 46 kDa cell-free translation product of a rat smooth muscle cell LOX mRNA, using an antibody directed against the 32 kDa bovine enzyme (Trackman et al. 1990, 1991), and a 48 kDa product, using mRNA isolated from fibrotic rat liver (Wakasaki & Ooshima 1990a). Three forms of LOX with molecular masses of 50, 45, and 32 kDa were identified by immunoprecipation of media and cell layer extracts of cultured rat smooth muscle cells, which were pulse-labelled with [35S]methionine (Trackman et al. 1992). The 50 kDa fragment was found both as a secreted protein in the cell medium and in the intracellular fraction, and pulse-chase studies revealed that it is converted to a 32 kDa protein in the medium. This 50 kDa fragment was found to be an N-glycosylated derivative of the 45 kDa proprotein (Trackman et al. 1992). Procollagen C-proteinase cleaves the 50 kDa precursor between Gly-168 and Asp-169 (numbered according to the human sequence) to yield the non-glycosylated 32 kDa mature enzyme (Figure 3) (Cronshaw et al. 1995, Panchenko et al. 1996). In mammals, procollagen C-proteinase activity is provided by products of the BMP1 gene, which encodes the alternatively spliced mRNA species for bone morphogenetic protein 1 (BMP-1), and mammalian tolloid (mTLD) (Takahara et al. 1994). The BMP1 gene is a member of a multigene family including two genetically distinct mammalian tolloid-related proteinases, tolloid-like 1 (mTLL-1) and tolloid-like 2 (mTLL-2). BMP-1, mTLD, and mTLL-1 all have procollagen C-proteinase activity, however, their substrate specificites vary. Recently, Uzel et al. (2001) demonstrated that BMP-1, mTLD, mTLL-1, and mTLL-2 all process the 50 kDa LOX precursor and control its activation in mouse embryo fibroblasts. In vitro assays with purified recombinant enzymes showed that all four proteinases productively cleaved the LOX precursor at the correct physiological site. However, BMP-1 was 3-, 15-, and 20-fold more efficient than mTLL-1, mTLL-2, and mTLD, respectively (Uzel et al. 2001). Fibroblasts from Bmp1 and Tll1 null mice produced both the 50 kDa LOX precursor and the processed LOX enzyme at similar levels, indicating apparently normal processing. In contrast, double null Bmp1/Tll1 cells produced predominantly the unprocessed 50 kDa precursor and had a 70% reduction in lysyl oxidase activity when compared to wild-type, Bmp1-null and Tll1-null fibroblasts. These results suggest that BMP-1/mTLD and mTLL-1 are responsible for the majority of LOX precursor processing into the active LOX enzyme, at least in murine embryonic fibroblasts. In addition, BMP-1 may be able to regulate the LOX activity through the processing event (Uzel et al. 2000).
2.3.2. Catalytic properties
2.3.2.1. Cofactors
Direct evidence that copper is a component of LOX was obtained from enzymes isolated from chick bone (Iguchi & Sano 1985, Iguchi et al. 1990), chick aorta (Harris et al. 1974), and bovine aorta (Gacheru et al. 1990). Atomic absorption spectroscopic studies have demonstrated that approximately one tightly bound copper atom is present in the 32 kDa LOX monomer. Removal of the copper ion leads to a catalytically inactive apoenzyme. Electron spin resonance studies have indicated that the copper in the resting enzyme is in the Cu(II) state, and is bound in a tetragonally distorted, octahedrally coordinated ligand field (Gacheru et al. 1990). The sequence WEWHSCHQHYH in human has been suggested to be the actual copper-binding region, which provides four histidine residues that are involved in the copper-binding coordination complex (Krebs & Krawetz 1993). In addition to the tightly bound Cu ion, purified LOX preparations have been shown to contain 5-9 atoms of loosely bound copper per enzyme molecule (Gacheru et al. 1990). Experiments in a cell free transcription/translation system in vitro have shown that the unprocessed 50 kDa LOX precursor binds copper (Kosonen et al. 1997). Further information of a post-translational incorporation of copper into LOX was obtained using skin fibroblasts incubated with different inhibitors of post-translational modifications, together with carrier-free 67Cu. When protein synthesis was inhibited with cycloheximide, only minor amounts of copper (bound to LOX) were found in the medium of cultured skin fibroblasts, indicating the importance of protein synthesis for the incorporation of copper into the enzyme. In contrast, inhibition of N-linked glycosylation with tunicamycin had no effect on secretion of the copper that was bound to LOX. Inhibition of processing of the 50 kDa LOX precursor into its 32 kDa active form with a procollagen C-proteinase inhibitor had no apparent effect on the amount of copper that was bound to LOX in the cell media (Kosonen et al. 1997). This observation was in agreement with earlier data by Kagan et al. (1995), which demonstrated that the propeptide region is not essential to the folding and secretion of the functional enzyme. However, when the assembly of secretory vesicles was inhibited with brefeldin A, the secretion of copper bound to LOX was inhibited (Kosonen et al. 1997), thus indicating that copper is incorporated into lysyl oxidase in the endoplastic reticulum or during protein trafficking through the Golgi elements (see Figure 3) (Kosonen et al. 1997, Rucker et al. 1998).
In additon to copper, LOX was long known to contain a covalently bound carbonyl prosthetic group (Levene 1961, Williamson et al 1986a, Williamson et al. 1986b, see Kagan 1986, 1994, Smith-Mungo & Kagan 1998 for reviews). However, the nature of this carbonyl group remained unknown for several decades. Knowledge obtained from other members of copper amine oxidases and their cofactors played an essential role in studies leading to the identification of the carbonyl cofactor (Klinman 1996). The cofactor of one of these family members similar to LOX, namely bovine plasma amine oxidase, was initially identified as pyrrolequinoline (PQQ) (Lobenstein-Verbeek et al. 1984). Comparison of the resonance Raman spectra of a proteolytic peptide obtained from LOX with the corresbonding spectrum of bovine plasma amine oxidase suggested that the cofactor in lysyl oxidase resembles PQQ (Williamson et al 1986b). However, strong evidence was presented later, indicating that the true carbonyl cofactor in bovine plasma amine oxidase is in fact topa quinone (TPQ) (Janes et al. 1990). Because TPQ is found in a wide range of copper amine oxidases, the presence of TPQ in LOX was regarded as plausible, but the large size difference between LOX and other amine oxidases (Klinman & Mu 1994), together with the absence of the consensus sequence for TPQ in LOX (Janes et al. 1990), raised the possibility of a different cofactor. This different and previously unknown cofactor was indeed found to occupy the active site of bovine LOX (Wang et al. 1996, Wang et al. 1997). Its structure is derived from the cross-linking of the ε amino group of a peptidyl lysine (Lys-314 in bovine LOX) with the modified side chain of a tyrosyl residue (Tyr-349 in bovine LOX), and it has been named lysine tyrosylquinone (LTQ). LTQ could be formed in the endoplasmic reticulum or during protein trafficking through the Golgi elements (see Figure 3) (Rucker et al. 1998). In this pathway, copper may play a structural role in stabilizing the LTQ (Tang & Klinman 2001).
2.3.2.2. Reaction mechanism
LOX catalyses primary amine oxidation through a ping pong bi ter kinetic mechanism (Williamson & Kagan 1986), which is illustrated in Figure 4. The first step in this reaction is the formation of a Schiff base with the lysine tyrosylquinone (LTQ) cofactor (I → II). While the LTQ is bound to the substrate, it undergoes a rate-limiting, general base-facilitated α proton abstraction (Williamson & Kagan 1987a). A histidine residue has been suggested to act as the general base in LOX (Gacheru et al. 1988). In the following reaction, stereospecific abstraction of the pro-S α-proton takes place (Shah et al. 1993a), and the electrons migrating from the substrate carbanion reduce the LTQ cofactor (II → III). Hydrolysis of the imine intermediate releases the reactive aldehyde product, which can then react spontaneously to form lysine- or hydroxylysine-derived cross-links. After the release of the aldehyde product, the reduced enzyme is reoxidized by molecular oxygen with the aid of Cu(II) to produce hydrogen peroxide and ammonia (Akagawa & Suyama 2001). By this way, the oxidized enzyme is regenerated and the catalytic cycle is completed (IV → V → I) (Kagan 1994, Smith-Mungo & Kagan 1998, Akagawa & Suyama 2001).
2.3.2.3. Inhibitors
Administration of β -aminopropionitrile (β APN) to growing animals results in a molecular disease known as lathyrism, which is characterized by an increased fragility of all connective tissues and an elevation of solubility of collagen from tissues due to diminished cross-linking (see Section 2.5.3). Therefore, the ability of β APN to inhibit cross-link formation was recognized even before identification of lysyl oxidase activity in vitro (Kagan 1986). β APN was later found to be a potent irreversible LOX inhibitor with a Ki of about 3-5 M (Narayanan et al. 1972). β APN binds to the active site of LOX and is enzymatically processed to a reactive form, which derivatizes the enzyme (Tang et al. 1983). β APN has therefore been used to specifically inhibit activities of all lysyl oxidase isoenzymes in numerous studies. The structure of β APN was used in a directed search for other inhibitors. β -Haloethylamines and β -nitroethylamine were identified as additional active site-directed irreversible inhibitors, with inhibition constants similar to that of β APN (Tang et al. 1984, Williamson & Kagan 1987a). Benzylamines substituted in the para position with electron-withdrawing functions behave as ground-state inhibitors, presumably by forming LOX enzyme-bound intermediates that are not completely processed into the aldehyde (Williamsson & Kagan 1987b). Vicinal diamines like cis-1,2-diaminocyclohexane and ethylenediamine are also potent irreversible LOX inhibitors (Gacheru et al. 1989). LOX is further inhibited by heparin (Gavriel & Kagan 1988), N-(5-aminopentyl)aziridine (Nagan et al. 1998), and trans-2-phenylcyclopropylamine (Shah et al. 1993b) with the latter functioning as a noncompetetive, reversible inhibitor (Shah et al. 1993b). LOX is additionally inhibited by homocysteine thiolactone and its selenium and oxygen analogs in an active-site-directed and irreversible manner (Liu et al. 1997a).
2.3.3. Regulation of LOX
2.3.3.1. Expression in different cell types and during development
LOX is expressed in several different cells, including fibroblasts, aortic and lung smooth muscle cells, osteoblasts, osteosarcoma cells (Csiszar et al. 1996, Uzel et al. 2000), myofibroblasts (Peyrol et al. 1997), corneal endothelial cells (Fujimaki et al. 1999), and chondrocytes (Gregory et al. 1999). Moreover, a secreted form of LOX is associated with tracheal chondrocytes, endothelial cells, basal cells, biliary epithelial cells, liver parenchymal cells, and spleen reticulum cells (Di Donato et al. 1997a). In kidney, LOX is detected in glomeruli, medulla, renal cell lines, and tubular epithelial cells (Di Donato et al. 1997a). At the tissue level, LOX is abundantly expressed and immunologically detected in fetal and adult aorta (Kagan et al. 1986, Baccarani-Contri et al. 1989), human placenta, skin, and lung (Baccarani-Contri et al. 1989).
During the development of different organisms, most observations of LOX expression patterns have been made in association with the assembly of collagen and elastin fibers in different tissues. With the aid of polyclonal antibodies, LOX has been localized in association with collagen fibers despite the age of the subjects, at least in skin and aorta. In contrast, only small amounts of aortic elastin fiber-associated LOX was observed in a 24-week-old human fetus, and by the age of 16 weeks, positive signals were completely diminished (Baccarani-Contri et al. 1989). In human amnion, the highest level of lysyl oxidase activity, as well the highest expression of LOX, were detected in gestational weeks 12-14, and the lowest values were seen at weeks 20-24 (Casey & MacDonald 1997). In adulthood, elastin fiber-associated LOX staining is mostly negative (Baccarani-Contri et al. 1989).
Between days 8 and 16 of chick embryonic development, the steady-state levels of LOX mRNA are increased together with its substrates, tropoelastin, and type I collagen. In situ hybridizations showed that the spatial expressions of the LOX and tropoelastin transcripts differ, suggesting that the enzyme and substrate genes may be regulated differentially within cells of the arterial wall (Wu et al. 1992). In a similar manner, the LOX mRNA level exhibited a fourfold increase from gestational day 9 to day 15 in rat embryos, the highest expression level being seen in day 13 (Tchaparian et al. 2000). However, the corresponding enzyme activity level was found to be relatively constant during those days. A measurable enzyme activity level was observed in rat embryos at gestational day 9, which corresponds to the transition of the embryo from the postblastocyst stage to preembryonic stage (Tchaparian et al. 2000). Moreover, in the embryonic development of rat, a pattern of temporal expression for a copper transporting P-type ATPase gene (ATP7A) was consistent with its possible role in copper delivery to LOX (Tchaparian et al. 2000).
During the development of sea urchin embryos, lysyl oxidase activity is increased severalfold, the activity being peaked during gastrulation and larva formation. Treatment of developing sea urchin embryos with β APN resulted in developmental arrest at the mesenchymal blastula stage, suggesting a critical role for LOX in mesenchyme migration, gastrulation, and morphogenesis during development (Butler et al. 1987). However, at the time of this study, it was supposed that only one enzyme is responsible for the cross-linking of elastin and collagens, and therefore affected by β APN (see Section 2.4 for details). Di Donato et al. (1997b) studied the role of LOX in Xenopus laevis oocytes, which are useful models in unraveling the signal transduction involved in mitogenesis and other oncogene-dependent processes. After the coinjection of LOX and the oncogenic p21-Ha-rasval12 into maturing oocytes, the ras-dependent maturation of these oocytes was inhibited as an effect of the intracellular LOX. Treatment of these coinjected oocytes with β APN abolished this inhibition of maturation. Results from this study suggest that the LOX protein is actually able to antagonize oncogenic Ras-dependent meiotic maturation of Xenopus laevis oocytes.
2.3.3.2. LOX in fibrotic tissues
Fibrosis is characterized by an extensive accumulation of the essentially insoluble collagen fibers. Therefore, it should be possible to treat this condition by selective suppression of key events in collagen biosynthesis, including the reaction catalyzed by the lysyl oxidases. Inhibition of lysyl oxidases would likely decrease the amount of cross-links in collagens, which may enhance its degradation by proteases (see Kagan 1986, 1994, 2000, Franklin 1995, 1997, Myllyharju & Kivirikko 2001 for reviews).
In the past decades, several reports have suggested a strong association between organ fibrosis and increased lysyl oxidase activity. Such observations have been made in experimental hepatic fibrosis in rat (Siegel et al. 1978, Carter et al. 1982, Wakasaki & Ooshima 1990a), in models of lung (Counts et al. 1981, Riley et al. 1982, Almassian et al. 1991), arterial (Kagan et al. 1981) and dermal fibrosis (Chanoki et al. 1995), in chronic human liver fibrosis (Murawaki et al. 1991), adriamycin-induced kidney fibrosis in rat (Di Donato et al. 1997a), and in other pathological conditions leading to fibrosis (Sommer et al. 1993, Jourdan-Le Saux et al. 1994, Ma et al. 1995, Decitre et al. 1998, Trivedy et al. 1999). Among these, probably the most striking increases in enzyme activity are seen in the rat model of CCl4-induced liver fibrosis (Siegel et al. 1978, Carter et al. 1982). In this model, an increase of lysyl oxidase activity reflects increases in the steady-state levels of LOX mRNA (Wakasaki & Ooshima 1990a). A significantly higher level of lysyl oxidase activity was also seen in the media of mesenchymal cells cultured from cirrhotic human livers when compared to that in the media of cells from normal liver or from livers of patients with chronic hepatitis (Konishi et al. 1985). In these studies, the low level of enzyme activity in the healthy liver increased 15- to 30-fold in fibrotic livers. Because the lysyl oxidase activity level is normally negligible in the serum of healthy subjects, but significantly increased in experimental liver fibrosis, it has been suggested that lysyl oxidase might serve as a marker of internal fibrosis (Siegel et al. 1978, McPhie 1981, Carter et al. 1982, Sakamoto et al. 1987, Murawaki et al. 1991).
Interestingly, changes in the expression level of type III collagen often precede or parallel changes in the level of lysyl oxidase activity (Krzyzosiak et al. 1992, Sharma et al. 1997, Kim et al. 1999). To clarify these findings, Giampuzzi et al. (2001) investigated the effects of LOX overexpression on the activity of the human COL3A1 gene promoter in monkey renal fibroblasts (COS-7) and human primary skin fibroblasts. Overexpression of the recombinant LOX led to a dramatic increase in the COL3A1 promoter activity. The data further suggested that induction of the COL3A1 promoter was dependent on the catalytic activity of LOX, as it was completely suppressed by β APN. Moreover, a sequence similar to the nuclear regulatory element 1 (NRE1) was found in the COL3A1 promoter and was shown to be able to bind Ku antigen, which is known to be involved in some of the main DNA repair and recombination processes (Jin & Weaver 1997, Jeggo 1998, Jeggo et al. 1999). However, the authors of this study had great difficulties in explaining and defining any possible mechanism by which LOX increased the Ku binding into its target sequence on the COL3A1 promoter (Giampuzzi et al. 2001).
In addition to the association between the expression of LOX and type III collagen, LOX expression also seems to be associated with type I collagen expression. In developing granulomas, LOX was transiently up-regulated at the transcriptional level in parallel with the α1 chain of type I procollagen (Sommer et al. 1993). Jourdan-Le Saux et al. (1997) subsequently demonstrated by transfection studies that the LOX and COL1A1 promoters may be regulated by similar negative and positive cis-acting elements. These regulatory regions include a potential TGF-β response element, which was also found in the rat Col1a1 (Ritzenthaler et al. 1991) and mouse Col1a2 (Rossi et al. 1988) promoters.
2.3.3.3. Response to different effectors in various cell types
LOX expression is highly responsive to a variety of physiological states, such as growth, wound repair, ageing, genetic diseases involving altered copper metabolism, and tumorigenesis. Therefore, it is obvious that expression of LOX is regulated by several different specific cytokines, growth factors, and related intercellular molecular messengers (see Smith-Mungo & Kagan 1998, Csiszar 2001 for reviews). The factors contributing to LOX expression and its activity in different cell lines, tissues, and physiological situations, and in several disorders, include specific transcription factors like the interferon response factor-1 (IRF-1) (Tan et al. 1996), metal ions (Li et al. 1995), cytokines and growth factors, such as transforming growth factor-β 1 (TGF-β 1) (Shibanuma et al. 1993, Boak et al. 1994, Feres-Filho et al. 1995, Roy et al. 1996, Gacheru et al. 1997, Hong et al. 1999, Shanley et al. 1997, Bose et al. 2000), basic fibroblast growth factor (bFGF) (Feres-Filho et al. 1996), fibroblast growth factor-2 (FGF-2), insulin-like growth factor (IGF) (Trackman et al. 1998), and platelet-derived growth factor (PDGF) (Green et al. 1995), hormones, such as testosterone (Bronson et al. 1987) and prostaglandin E2 (PGE2) (Boak et al. 1994, Roy et al. 1994, Choung et al. 1998), and signaling molecules like ras (Contente et al. 1990, Hajnal et al. 1993, Kryzosiak et al. 1992, Csiszar et al. 1996, Di Donato et al. 1997b) and cAMP (Choung et al. 1998, Ravid et al. 1999). The growth factors and other effectors that are known to affect LOX expression and activity are listed in Table 1. Because of cell type differences and experiments carried out under a variety of conditions, it is difficult to draw any general conclusions concerning the transcriptional and post-translational regulation of LOX.
One of these effectors, TGF-β 1, is a fibrogenic growth factor known to regulate the synthesis of collagens (Chambers & Laurent 1996, Wight 1996) and elastin (Cleary & Gibson 1996). TGF-β 1 has been found to strongly promote expression of LOX in fibroblasts from neonatal rat lungs (Boak et al. 1994) and human embryos (Roy et al. 1996), and in rat vascular smooth muscle cells (Gacheru et al. 1997, Shanley et al. 1997). In human lung fibroblasts, the increase of LOX expression induced by TGF-β 1 is effectively prevented by PGE2 (Roy et al. 1996). PGE2 is known to induce the expression of cyclooxygenase-1, the catalyst leading to the production of PGE2, and therefore it has been suggested that a coordinated and autocrine-like mechanism could limit LOX expression in inflammatory responses to connective tissue injury (Roy et al. 1996). Most of the regulatory effects of TGF-β 1 on LOX expression in cultured rat aorta smooth muscle cells probably occurs at a post-transcriptional level by an enhancement of the stabilization of the LOX mRNA (Gacheru et al. 1997). In MC3T3-E1 osteoblastic cells, TGF-β 1 also stimulated LOX expression both at the mRNA and protein levels. However, proteolytic prosessing of the LOX precursor was proportionately less stimulated by TGF-β 1, which may account for the delay and slightly lower magnitude of the increase in the corresponding enzyme activity when compared to the peak of the increase in the mRNA level (Feres-Filho et al. 1995). TGF-β 1 also regulates BMP-1 and mTLD (Lee et al. 1997), which are responsible for the proteolytic processing of the LOX precursor (Uzel et al. 2001; see also Section 2.3.1.1), and therefore serves as one of the possible regulatory levels in the activation of the enzyme. In contrast, this processing event did not seem to be affected by the elevation of LOX expression by TGF-β 1 in cultured neonatal rat lung fibroblasts (Boak et al. 1994).
In adult rat vascular smooth muscle cells, the LOX mRNA level and the secreted enzyme activity were low in quiescent cells from adult rats, but were markedly induced by PDGF, angiotensin II, and by increasing the serum concentration to 10 % (Green et al. 1995). However, in quiescent neonatal rat vascular smooth muscle cells, the expression of LOX mRNA and the secreted enzyme activity were elevated, but were markedly decreased by stimulation of the proliferation by serum enrichment (Gacheru et al. 1997), suggesting age-specific mechanisms of regulation in these cells. Interferon ã (IFN- ã) has been shown to downregulate LOX gene expression in rat aortic smooth muscle cells by transcriptional and posttranscriptional mechanisms (Song et al. 2000). This agrees with the finding that an IFN-sensitive motif is present in the promoter of the LOX gene (Tan et al. 1996).
Table 1. Transcriptional and posttranscriptional regulation of LOX.
| Effector | Cell or tissue | Effect |
|---|---|---|
| IFN-γ 1 | Rat aortic smooth muscle cells | Downregulation of mRNA; decreased mRNA half-life |
| bFGF2 | Mouse osteoblastic cells | Decreased mRNA level (1-10 nM); upregulation of mRNA (0.01-0.1 nM) |
| FGF-2 and IGF-13 | Inflamed oral tissues, rat fibroblastic mesenchymal cells | Increase of mRNA |
| PGE24-6 | Rat lung fibroblasts | Unchanged mRNA level; reduction in enzyme activity |
| Human embryonic lung fibroblasts | Downregulation of mRNA | |
| TGF-β 14,5,7-12 | Rat aortic smooth muscle cells | Increased mRNA level and enzyme activity |
| Human gingival, flexor reticulum cells, renal cell lines | Increased mRNA level and enzyme activity | |
| Murine osteoblast-like cells | Increased mRNA level and enzyme activity | |
| Cadmium13 | Mouse fibroblasts | Decreased mRNA level |
| Mouse cadmium-resistant fibroblasts | Increased mRNA level | |
| Testosterone14 | Calf aortic smooth muscle cells | Increased enzyme activity |
| Bleomycin15 | Human lung fibroblasts | Increased mRNA level |
| Human dermal fibroblasts | Decreased mRNA level | |
| Hydralazine15 | Human dermal fibroblasts | Increased mRNA level |
| Minoxidil15 | Human dermal fibroblasts | Increased mRNA level |
| Adriamycin16 | Rat kidney glomeruli, medula | Increased mRNA level |
| cAMP6,17 | Rat and human vascular smooth muscle cells | Upregulation of transcription |
| PDGF18 | Rat vascular smooth muscle cells | Upregulation of mRNA |
| Dexamethasone19 | Cultured fetal murine lungs | Upregulation of mRNA |
| Retinoic acid20 | Adipocytes in early adipogenesis | Prevents downregulation of mRNA and enzyme activity |
| References: 1Song et al. 2000, 2Feres-Filho et al. 1996, 3Trackman et al. 1998, 5Roy et al. 1996, 4Boak et al. 1994, 6Choung et al. 1998, 7Shibanuma et al. 1993, 8Feres-Filho et al. 1995, 9Gacheru et al. 1997, 10Hong et al. 1999, 11Shanley et al. 1997, 12Bose et al. 2000, 13Li et al. 1995, 14Bronson et al. 1987. 15Yeowell et al. 1994, 16Di Donato et al. 1997a, 17Ravid et al. 1999, 18Green et al. 1995, 19Chinoy et al. 2000, 20Dimaculangan et al. 1994. | ||
2.3.4. Novel biological roles
LOX has been reported not only to be involved in the cross-linking of collagens and elastin, but to also act as a tumor suppressor and to have intracellular and even intranuclear activities (for reviews, see Smith-Mungo & Kagan 1998 and Csiszar 2001).
2.3.4.1. LOX in tumor suppression
The first direct evidence of the tumor suppressor activity of LOX was demonstrated by Contente et al. (1990). They identified a markedly down-regulated cDNA species upon transformation of mouse NIH 3T3 cells with LTR-c-H-ras. The expression level of this cDNA was restored when the transformed cell line (RS485) was treated with interferon to obtain a persistent revertant cell line (PR4). This cDNA species was named as ras recision gene (rrg). These persistent revertant cells are phenotypically non-transformed and non-tumorgenic. However, when a rrg antisense RNA was used to block the expression of the corresponding mRNA in PR4 cells, the reappearance of the tumorgenic and transformed phenotype was observed. Furthermore, subcutaneous injection of these antisense cells into athymic mice induced tumor formation. The cDNA sequences of the mouse rrg and rat Lox were subsequently found to be identical, thus indicating that rrg encodes Lox (Kenyon et al. 1991), the activity of which was already known to be markedly low in the medium of malignantly transformed cultured human cell lines (Kuivaniemi et al. 1986b). These results have been supported by several other groups. Expression of LOX was down-regulated in immortalized rat 208F fibroblasts after transformation by activated H-ras (Hajnal et al. 1993, Oberhuber et al. 1995) and up-regulated in spontaneous phenotypic revertants that continued to express the ras oncogene (Hajnal et al. 1993). In addition, LOX expression was found to be irreversibly and coordinately regulated with type I collagen expression in spontaneous phenotypic revertants (Hajnal et al. 1993). A putative tumor suppressor role for LOX was also demonstrated by stable transfection of a LOX cDNA in antisense orientation to normal rat kidney fibroblasts (NRK-49F). The transfected cells exhibited a loose attachment to plate, anchorage-independent growth, and high tumorgenicity in nude mice, and also showed an impaired response of the PDGF and IGF-1 receptors to their ligands (Giampuzzi et al. 2001). Hämäläinen et al. (1995) demonstrated that the low lysyl oxidase activity level in several malignantly transformed human cell lines is due to low quantities of the LOX mRNA and a low transcription level of the corresponding gene. A similar transcriptional silencing of LOX expression was detected in H-ras-transformed rat embryo fibroblasts (CREF) by using nuclear run-on assays (Su et al. 1995). The LOX gene was also identified as a target for the anti-oncogenic transcription factor (IRF-1), which manifests tumor suppressor activity and contributes to the development of human hematopoietic malignancies (Tan et al. 1996).
In malignant human breast carcinomas, LOX was highly expressed in myofibroblasts and myoepithelial cells surrounding the in situ tumor and in the reactive fibrosis facing the invasion front of infiltrating tumors (Peyrol et al. 1997). Type I, III, and IV collagens and elastin were found to be co-distributed with LOX in this study, thus resulting in the stabilization of a scar-like peritumor barrier. In contrast, LOX was found to be absent from the carcinoma cells. These findings suggested a possible host defense mechanism, while a late stromal reaction lacking LOX favors tumor dispersion (Peyrol et al. 1997).
A similar phenomenon was observed in experimental mouse prostate cancer, where LOX was expressed in normal epithelium, but progressively lost in primary prostate cancer and associated metastatic lesions (Ren et al. 1998). In the stromal reaction of different types of broncho-pulmonary carcinoma, a strong expression of LOX is associated with the hypertrophic scar-like stromal reaction found at the front of tumor progression. In contrast, little or no expression was found within the stromal reaction of invasive carcinomas, small cell carcinomas, and neuro-endocrine carcinomas (Peyrol et al. 2000). Very recently, it was demonstrated that a loss or reduction of LOX function during tumor development may be a direct consequence of somatic mutations and may be associated with the pathogenesis of colon cancer (Csiszar et al. 2002). The LOX gene is located in chromosomal region 5q23 (Hämäläinen et al. 1991, 1993, Mariani et al. 1992, Boyd et al. 1995), which is known to be deleted at a high frequency in many different types of cancer (Tamura et al. 1996, Wieland et al. 1996).
2.3.4.2. Intracellular and intranuclear activities
LOX has been localized within fibroblasts, chondrocytes, smooth muscle cells, and a variety of non-fibroblastic cells. In cultured fibroblasts, LOX was immunologically detected in association with filamentous structures in the cytoplasm consistent with cytoskeletal proteins (Wakasaki & Ooshima 1990b). However, since the molecular weight of this intracellularly observed enzyme was not assessed, it was unknown wheter it represented the proenzyme and/or mature LOX form. It was recently suggested that an intracellularly expressed recombinant mature 32 kDa LOX form may be able to regulate the activity of the COL3A1 gene promoter, as discussed in Section 2.3.3.2 (Giampuzzi et al. 2001). In addition, coinjection of LOX with oncogenic p21-rasval12 into Xenopus laevis oocytes suggests one possible intracellular role for LOX in antagonizing a Ras-induced meiotic maturation of these cells. The authors of this study suggested that a LOX-dependent block in oocyte maturation may be downstream of Erk2, a member of the mitogen-activated protein kinases (Di Donato et al. 1997b).
In addition, a 32 kDa active form of LOX has been localized by immunocytochemistry and Western blot analysis to the nucleus of rat vascular smooth muscle cells and 3T3 fibroblasts (Li et al. 1997). Moreover, the nucleus of the vascular smooth muscle cells contained lysinonorleucine, which is the adduct formed during the LOX catalyzed cross-link reaction. Formation of lysinonorleucine was prevented by administration of β APN, thus confirming a role of LOX in this reaction (Li et al. 1997). More recently, it was demonstrated by using cultured smooth muscle cells that a purified bovine 32 kDa LOX polypeptide fluorescently labeled with rhodamine was able to enter into the cytosol and become rapidly concentrated into the nuclei of these cells. The intracellular uptake and distribution were not altered by β APN, and therefore were independent of the catalytic activity of LOX (Nellaiappan et al. 2000). The identity of a nuclear substrate of LOX is currently unknown (Li et al. 1997).