Chapter 6. Discussion
6.1. Cloning and characterization of the fourth and the fifth lysyl oxidase isoenzymes (I, II)
6.1.1. Primary structures of the LOXL3 and LOXL4 polypeptides (I, II)
We identified two additional human lysyl oxidase isoenzymes; lysyl oxidase-like 3 protein (LOXL3) and lysyl oxidase-like 4 protein (LOXL4). Our findings were largely supported by Huang et al. (2001) and Jourdan-Le Saux et al. (2001), concerning the characterization of LOXL3 isoenzyme, and by Asuncion et al. (2001), concerning the characterization of LOXL4 isoenzyme. Furthermore, the mouse lysyl oxidase-related protein 2 identified by Jang et al. (1999) was found to be orthologous to human LOXL3 (Huang et al. 2001, Jourdan-Le Saux et al. 2001).
The LOXL3 and LOXL4 polypeptides have a high overall similarity to each other, but also to the LOXL2 polypeptide, and each of them has four conserved SRCR domains at their N-terminal region (see Figure 5, and Figure 1 in II). LOXL3 and LOXL4, however, share a significant similarity to isoenzymes LOX and LOXL only in the C-terminal region, which contains the putative copper-binding region (Krebs & Krawetz 1993), the lysine and tyrosine residues that form the lysine tyrosylquinone cofactor (Wang et al. 1996), and a growth factor and cytokine binding domain (Bazan 1989, 1990, Kim et al. 1995). These features of the catalytic i.e. LO domain are highly conserved within all lysyl oxidase isoenzymes (Figure 5).
The existence of a putative signal sequence and four SRCR domains suggests that the LOXL3 and LOXL4 isoenzymes are extracellular proteins (Figure 5). SRCR domains have been reported to be able to interact with other proteins (Resnick et al. 1994, Yamada et al. 1998), and therefore LOXL3 and LOXL4 may be involved in the binding and cross-linking of cell surface and extracellular matrix proteins. Interestingly, LOXL2 has been suggested to be involved in cell adhesion based on its significantly higher expression levels in adherent than nonadherent cell lines (Saito et al. 1997). Based on these data, Saito et al. (1997) proposed that the LOXL2 function may involve posttranslational modification of extracellular matrix components or other cell membrane proteins. This would be also a plausible function for LOXL3 and LOXL4 because of their SRCR domains.
6.1.2. The structure of the LOXL3 gene, and expression of LOXL3 and LOXL4 mRNAs in various tissues (I, II)
According to the information gained from HTGS sequences, the LOXL3 gene has 14 exons, and it is located on chromosome 2p13. The gene encodes a single 3.1 kb mRNA identified in Northern blots from several tissues. The size of the mRNA corresponds well to the size of 2.8 kb for an mRNA lacking the poly(A)+ tail, assuming that the first of the three potential polyadenylation signals is used. Northern analyses indicated the highest expression levels of the LOXL3 mRNA in the placenta, heart, ovary, testis, small intestine, and spleen. Distinct hybridization signals were, however, obtained from all tissues examined.
Based on the HTGS sequence, the LOXL4 gene is located on chromosome 10. It codes for a single mRNA of approxymately 4.0 kb in size, which is expressed in most human tissues studied, the highest expression levels were seen in skeletal muscles, testis, and pancreas. No expression, however, was detected in the brain or blood leukocytes. Our findings disagree with those made by Ito et al. (2001) concerning the corresponding mouse gene, Loxc, that was reported to show a cartilage specific expression. Because the corresponding human gene is expressed in many different human tissues, it would seem appropriate to call this new polypeptide LOXL4 rather than LOXC.
6.1.3. Recombinant expression of the LOXL3 and LOXL4 polypeptides (I, II)
Western analyses of lysates from both the LOXL3- and LOXL4-transfected HT-1080 cells showed the expression of a 97 kDa polypeptide. This size is slightly larger than the estimated overall molecular masses of 83.6 kDa for LOXL3 and 84.5 kDa for LOXL4. The differences are probably due to the V5 epitope and the histidine tag located at the C-terminal end of the recombinant polypeptides. Analysis of concentrated medium samples showed the presence of at least two major LOXL3 polypeptides and at least one LOXL4 polypeptide slightly larger in size than the corresponding polypeptides present in the cell lysates, as well as a very minor 97 kDa polypeptide. Since both recombinant polypeptides have several putative glycosylation sites, it is possible that these differences in sizes are due to utilization of the glycosylation sites.
The 50 kDa precursor of human LOX is cleaved by procollagen C-proteinase between Gly-168 and Asp-169 to yield a non-glycosylated, 32 kDa mature enzyme (Cronshaw et al. 1995, Panchenko et al. 1996). Two putative Gly-Asp processing sites are found in the bovine LOXL polypeptide (Borel et al. 2001), while one such site is present in the human LOXL2 polypeptide (Jourdan-Le Saux et al. 1999), but the possible utilization of none of these sites has been demonstrated experimentally. The bovine LOXL protein was found to be largely inactive, but was activated by processing with procollagen C-proteinase, suggesting that one or both of these sites may be involved in the processing event (Borel et al. 2001). A similar potential cleavage site for procollagen C-proteinase was also found in the LOXL3 polypeptide. If proteolytic processing of the LOXL3 polypeptide occurred between Gly-447 and Asp-448, the predicted size of the cleaved product would be 306 amino acids, with a theoretical molecular mass of 34.8 kDa. However, no intracellular or extracellular processing of the LOXL3 polypeptide was detected during recombinant expression in the transfected HT-1080 cells or by the addition of conditioned medium of human primary skin fibroblasts (data not shown).
The consensus sequence for the procollagen C-proteinase cleavage site may not be highly conserved (see Borel et al. 2001 for details). Therefore, despite of the lack of any potential procollagen C-proteinase cleavage site in the LOXL4 polypeptide, this isoenzyme was also tested for possible processing by recombinant expression in HT-1080 and CHO cells and in mouse embryonic fibroblasts (data not shown). No intracellular or extracellular processing of the LOXL4 polypeptide was detected in these cells. The results do not, however, exlude the possibility that the nonrecombinant LOXL3 and LOXL4 polypeptides may be processed in some other cell types, and thus further studies are needed to exclude or demonstrate the processing event.
The N-terminal region of the LOXL3 polypeptide contains a bipartite nuclear localization signal (KKQQQSKPQGEARVRLKG), which is not found in any other lysyl oxidase isoenzyme. Despite the existence of the nuclear localization signal, our results suggest that LOXL3 may not contribute to the lysyl oxidase activity identified in the nucleus, which has been shown to be due at least in part to the LOX polypeptide (Li et al. 1997). The present data do not exclude the possibility that the nonrecombinant LOXL3 polypeptide may be found in the nuclei of cells of some other type, and thus further studies are needed to either exclude or identify a nuclear location.