VALORES Y FORMACION PROFESIONAL

1.3.5.1 Golgi p i,4 -GalTase

pi,4-GalTase has been localized predominantly to the trans Golgi where it was found to codistribute with thiamine pyrophosphatase (Roth and Berger, 1982; Strous, 1986; Ulrich et al., 1986; Watzele et al., 1991; Taatjes et al., 1992; Berger et al., 1993). However, in some cells the pl,4-GalTase distribution has been found to overlap into the trans Golgi network (Geuze et al., 1985; Taatjes et al., 1987; Nilsson et al., 1993b; Rabouille et al., 1995). The oligosaccharides on p 1,4-GalTase contain sialic acids

which also suggests that there is some overlap between the glycosyltransferases and/ or the biosynthesis of pi,4-GalTase involves some sort of retrieval trafficking from the trans Golgi network to the trans Golgi. In several cell types the colocalization of pl,4- GalTase and ST6Gal I to the trans Golgi cistemae could be distinguished into separate compartments following treatment with monensin and to a lesser extent with chloroquine, suggesting discrete subcompartmentation (Berger, et al., 1993). Chloroquine has been shown to inhibit sialylation, but not affect galactosylation, of immunoglobulins in chloroquine-treated plasma cells (Thorens and Vassalli, 1986). The ability of individual glycosyltransferases to concentrate within certain Golgi locales requires a retention mechanism(s) (Shaper and Shaper, 1992; Machamer, 1993; Pelham and Munro, 1993; Colley, 1997). All the cloned Golgi glycosyltransferases are type II membrane-bound proteins, yet share little sequence homology. Therefore the retention signals are most likely to be general conformations rather than exact peptide sequence.

At present there are two potential Golgi retention mechanisms being tested which may work together: 1) the bilayer thickness model, which proposes that the shorter transmembrane domain of Golgi proteins prevent them from entering the cholesterol- rich vesicles for targeting to the plasma membrane (Bretscher and Munro, 1993; Masibay et al., 1993; Munro, 1995). It is thought that a higher concentration of cholesterol exists across the Golgi stack (Orci et al., 1981) 2) The oligomerization/ kin recognition model, which proposes that the oligomerization of Golgi proteins blocks their entry into transport vesicles (Nilsson et al., 1993a; Gleeson et al., 1994; Rabouille, et al., 1995; Nilsson et al., 1996)

Golgi retention of ST6Gal I indicated that retention may be attributed to several domains including the transmembrane domains, flanking sequences and residues present within the stem region (Munro, 1991; Colley et al., 1992; Dahdal and Colley, 1993; Masibay, et al., 1993). Recently, it has been found that the stem region of N- acetylglucosaminyltransferase I (GnT I) was required and sufficient for kin recognition of mannosidase II and Golgi retention (Munro, 1995; Nilsson, et al., 1996). Studies with pl,4-GalTase deletion mutants and chimeras have identified the major role the transmembrane domain plays in anchoring pl,4-GalTase to the trans-Golgi (Nilsson et al., 1991; Aoki et al., 1992; Teasdale et al., 1992; Masibay, et al., 1993; Teasdale et al.,

1994; Yamaguchi and Fukuda, 1995).

Comparison of the sequences in the transmembrane domain and the area flanking the N- terminal side, show a very high degree of homology between mammalian and avian pl,4-GalTases. This conserved sequence may be relevant to Golgi retention (Shaper and Shaper, 1992). Point mutations of Cys^^ to Ser and His^^ to Leu within the

transmembrane domain of pl,4-GalTase resulted in the loss of some Golgi retention and appearance on the cell surface (Aoki, et al., 1992). Further, stable expression of pl,4- GalTase was sensitive to deletions in the transmembrane domain, as deletion of the first five amino acids, up to the cysteine residue, in this region abolished activity (Masibay, et al., 1993). However, Nilsson et al (1991) found that chimeric constructs using the human invariant chain as a reporter protein achieved Golgi retention with just the luminal half of the transmembrane domain of pl,4-GalTase (which excluded Cys^^ and His^2). Other Golgi proteins show no primary sequence homology between their transmembrane domains. Additional studies have indicated that pl,4-GalTase as well as other Golgi localized enzymes exist as dimers (Navaratnam, et al., 1988; Bendiak et al., 1993; Fleischer et al., 1993; Colley, 1997}. Overexpression of pi,4-GalTase- transferrin chimeras resulted in a small percentage of p-mercaptoethanol-resistant homo-dimers which required the Cys^^ and His^^ residues in the transmembrane domain (Yamaguchi and Fukuda, 1995). Even less dimer formation was apparent with the pl,4-GalTase wild-type constructs. However, full-length pi,4-GalTase stably transfected into murine L cells was found to exist almost entirely as high molecular weight oligomers, which may indicate a role in Golgi retention (Teasdale, et al., 1994). Purified soluble pl,4-GalTase from different sources has been shown to exist as a dimer (Powell and Brew, 1974; Malissard, et al., 1996). pl,4-GalTase isolated from the rat liver Golgi membranes was found to form large oligomers which were dependent upon a 66 N-terminal amino acid sequence (Bendiak, et al., 1993). This data would suggest that pl,4-GalTase has some inherent capacity to self-associate without the need for additional molecules. Although the intracellular dimer formation may well be different from that seen with the purified form, in that additional complexes with other Golgi proteins, lipid bilayer and cytoskeletal components are involved (Nilsson et al., 1994; Slusarewicz et al., 1994; Nilsson, et al., 1996). It has been found that the Golgi enzymes pl,4-GalTase, ST6Gal I, uridine diphosphatase and a-mannosidase II are all functionally active as dimers (Moreman et al., 1991; Fleischer, et al., 1993). However, another disulphide-bonded dimer of ST6Gal I, representing about 30% of the enzyme in rat liver, was found to be catalytically inactive yet could still bind galactose (Ma and Colley, 1996). Rat hepatoma cells stimulated with dexamethasone synthesized homodimers of ST6Gal I in vivo which constituted approximately 20 % of the total immunoprecipitated ST6Gal I from these cells (Bosshart and Berger, 1992).

The cytoplasmic tail may play an accessory role in the retention process (Nilsson, et al., 1991; Evans et al., 1995). Nilsson's work suggested that the cytoplasmic tail, provided either by p 1,4-GalTase or an isoform of the human invariant chain, Iip31, in combination with the transmembrane domain sustained complete retention of pl,4-

GalTase in the Golgi. In contrast, Shur and colleagues have attributed the role of the cytoplasmic tail in pl,4-GalTase to its cell surface expression.

1.3.5.2 Cell surface p i,4 -GalTase

pl,4-GalTase expression is also found on the plasma membrane of certain cell types using biochemical and immunodetection techniques. It is unlikely that cell surface pl,4-GalTase galactosylates structures in vivo as there is no evidence of sufficient concentrations of UDP-galactose found extracellularly. However, the pl,4-GalTase still has the capacity to bind GlcNAc terminating glycoconjugates and could therefore act like a lectin with other cell surface molecules or cell matrix molecules exposing the appropriate substrates (Shur, 1991). The functional roles of cell surface pi,4-GalTase have included the following possibilities: fertilization of murine ova (Miller et al., 1992); a receptor for polymeric IgA (Aicher et al., 1992); cell migration (Eckstein and Shur, 1989; Shur, 1993), developmental growth (Hinton et al., 1995); neurite outgrowth (Huang et al., 1995); and a signal transducing receptor (Gong et al., 1995).

Whist the detection of pl,4-GalTase at the cell surface is acknowledged, the molecular mechanism(s) responsible for this localization is contested. Very recently two similar liver ST6Gal I have been identified which differ by a single amino acid Cys/Tyr in the catalytic domain at position 123. Low levels of the ST6Gal I Tyr^^^ was found on the cell surface of transfected COS cells and high levels secreted into the tissue culture medium, whereas the ST6Gal I Cys^^^ was retained within the Golgi (Ma et al., 1997). Differential transcription of the pi,4-GalTase gene can result in a long- or short- pl,4- GalTase protein which are identical except for a 13 amino acid cytoplasmic tail extension in the long form (see section 1.2.2 and Figure 1.2). Most somatic cells were found to produce low levels of the long pl,4-GalTase, suggesting that the transmembrane domain is of prime importance for Golgi retention (Harduin^, et al.,

1993). Murine sperm only express the long pl,4-GalTase (Shaper, et al., 1990) which is localized to the plasma membrane (Lopez et al., 1985). Stable transfections of the long and short form pl,4-GalTase cDNA suggest that the long form pl,4-GalTase preferentially locates to the plasma membrane (Lopez et al., 1991), which has been confirmed in both F9 and 3T3 cells using antibodies against the 13 amino acid tail extension (Youakim et al., 1994a). Overexpression of the long form of pl,4-GalTase in COS-1 cells lead to increased protein in both the Golgi and the plasma membrane, whereas the short pi,4-GalTase only targeted to the Golgi (Dinter and Berger, 1995a). Variable cellular locations has been found in different cells expressing constructs at different levels (Teasdale, et al., 1994; Dinter and Berger, 1995a) (see chapter 5). Consistent with the cell surface localization of pl,4-GalTase was a post-translational modification, likely to have occurred in the trans Golgi network (Teasdale, et al., 1994). Further, it was thought unlikely that any of the cell surface pl,4-GalTase was recycled

to the Golgi (Teasdale, et al., 1994). Others have shown that overexpression of proteins can lead to retention in the ER but not saturate the Golgi and cause a leakage of the excess protein onto the cell surface (Nilsson, et al., 1991; Russo et al., 1992). Further, other groups have shown no preferential localization of the long form of pi,4-GalTase to the plasma membrane in stably transfected CHO cells (Russo, et al., 1992) or when highly expressed in COS-1 (Teasdale, et al., 1992), COS-7 cells (Masibay, et al., 1993), or HeLa cells (Nilsson, et al., 1991). Possible explanations for this conflicting data may be due in part to different experimental approaches such as the reporter proteins used in chimeras, variations in the transfection efficiencies, the detection methods used and the particular cell types which utilize different localization signals (Tang et al., 1995). A potential caveat in studying retention signals with chimeric molecules or mutants proteins may be the creation of retention signals or cause the subsequent loss of a transport mechanism (Low et al., 1994; Low et al., 1995; Wahlberg et al., 1995).

Masibay et al. (1993) found that increasing the number of hydrophobic residues (using isoleucines) in the transmembrane region of pl,4-GalTase led to an increase in its cell surface location on COS cells. Similar results have been seen for other Golgi proteins (Munro, 1991; Burke et al., 1994), though it was not an all-or-nothing process suggesting that other sequences or retention mechanisms are involved. There is some indication that both of the above models may be involved in localizing pi,4-GalTase to the trans Golgi. At present it is unclear if the disruption in Golgi retention is mainly due to an increased number of transmembrane residues or if the folding and recognition signal of amino acids has been perturbed. Replacement of the transmembrane domain in GnT I with leucines did not alter its kin recognition of mannosidase II or its Golgi retention but severely disrupted the Golgi structure (Nilsson, et al., 1996). Very recently two similar liver ST6Gal I have been identified which differ by a single amino acid Cys/Tyr in the catalytic domain at position 123. Low levels of the ST6Gal I Tyr^^^ was found on the cell surface of transfected COS cells and high levels secreted into the tissue culture medium, whereas the ST6Gal I Cys^^^ was retained within the Golgi (Ma, et al., 1997).

During the biosynthesis of pl,4-GalTase it undergoes palmitation, probably at or just before entry in the Golgi complex (Strous, 1986). Addition of fatty acid groups on other integral membrane proteins has been located near to or within the transmembrane region and linked to either a serine, threonine or cysteine residue (Strous, 1986; James and Olson, 1990}. Such residues exist in this region in pi,4-GalTase and it would be interesting to speculate if fatty acylation played a role in membrane targeting (Masibay, et al., 1993) and/or cell signalling pathways (Molenaar et al., 1988). Post-translational

phosphorylation of pl,4-GalTase may also participate in Golgi localization (Strous, 1986).