INTERESES PROFESIONALES

As the soluble form of pi,4-GalTase is catalytically active for both pl,4-GalTase and lactose synthetase activities, the cytoplasmic tail, transmembrane domain and stem region, up to amino acid 129 (Boeggeman et al., 1993; Wang et al., 1994), are not required for enzymatic activity per se. N-terminal deletion mutants of pl,4-GalTase showed very similar apparent Km values for GlcNAc, chitobiose and chitotriose as those for bovine milk pi,4-GalTase (Boeggeman et al., 1995). The essential binding sites for the donor, acceptor and Mn^+ must therefore reside within the soluble form of pl,4-GalTase. Chemical modifications of pl,4-GalTase has identified key amino acid residues involved in the binding of substrates. Ultraviolet photo-inactivation of pl,4- GalTase, and inactivation with 2-hydroxy-5-nitrobenzyl bromide, revealed the importance of tryptophan in the pl,4-GalTase—Mn^+—UDP-galactose interaction (Clymer et al., 1976). Aromatic residues were also shown to be involved in the same pl,4-GalTase—Mn^+—UDP-galactose interaction complex as detected using circular dichroism (Geren et al., 1975). Incorporation of 125% into pi,4-GalTase resulted in modification of tyrosines with loss in enzyme activity to various degrees. Tyrosines could be protected by a-lactalbumin or UDP-galactose, however, combinations including a-lactalbumin, UDP and GlcNAc afforded the greatest protection (Silvia and Ebner, 1980). Further differential spectroscopic studies have also shown the importance of tyrosine and tryptophan in the pi,4-GalTase—Mn^+—UDP-galactose interaction (Takase and Ebner, 1981). Chemical modification with a periodate-cleaved UDP and trace acétylation analysis, identified Lys^"^^, located next to one of the seven cysteines in pi,4-GalTase, and Lys^^l the binding of UDP-galactose (Yadav and Brew, 1990). Even though cysteine residues are important in pl,4-GalTase enzymatic function as demonstrated by inhibition with a variety of sulphydryl reagents, the inhibition was not complete as approximately 5-15% of the original activity remained (Magee and Ebner, 1974a; Wong and Wong, 1984). The binding sites on bovine milk pl,4-GalTase for a - lactalbumin and UDP-galactose appear to be separate as suggested by photo-affinity labelling with a UDP-galactose analogue (Lee et al., 1983). The a-lactalbumin interacts in the N-terminal portion of the molecule between residues 93 and 250 (Yadav and Brew, 1991), whilst UDP-galactose binds in the carboxyl end of pl,4-GalTase from amino acid 275 (Yadav and Brew, 1990).

Since the availability of the pl,4-GalTase cDNA, recombinant DNA technologies have been used in structure and function studies. Boeggeman et al (1995) have localized the binding sites on pl,4-GalTase for GlcNAc and UDP-galactose to distinct domains. Bovine cDNA mutants were expressed in E. coli and binding sites for GlcNAc and UDP-galactose (via Mn^+) were found to occur in the N-terminal (residues 130-257) and C-terminal (residues 258-402) regions respectively (Boeggeman, et al., 1995). GlcNAc was also found to bind to the C-terminal fragment, but only the interaction with the N-terminal fragment exhibited similar binding characteristics to catalytically active pi,4-GalTase protein (residues 130-402). However, using human cDNA mutants obtained through site-directed mutagenesis and expressed in E. coli, substrate binding sites were found to be in a much closer proximity to each other. The binding area centred around a hydrophobic pocket and included residues Tyr^^^, Tyr^^^ and Trp^^^ involved in GlcNAc binding with Tyr^^9 also involved in the binding of UDP- galactose (Aoki, et al., 1990). Further studies around this hydrophobic area identified Phe305pro306Asn^®^Asn^®^ residues involved in the UDP-galactose binding site, though they did not affect the binding of Mn^+ (Zu et al., 1995). The apparently contradictory data on substrate binding sites in pi,4-GalTase may be explained if the catalytic domain of p 1,4-GalTase was accessible enough to allow weak binding of substrates. The interaction of UDP-galactose and acceptor substrate may then induce a conformational change which would allow a very close association between the substrate binding regions to permit the transfer of galactose and the formation of the glycosidic bond. Hence the binding of substrates on separate pi,4-GalTase fragments containing N- terminal (130-257) or C-terminal (258-402) residues, would not cause such a conformational change. Similarly, if substrate analogues were able to bind to the substrate binding sites on pl,4-GalTase but unable to induce the required conformational change, then no pl,4-GalTase reaction would occur. It has been suggested that pi,4-GalTase may have two GlcNAc binding sites. Synthetic divalent spacer substrates, containing two molecules of GlcNAc separated by different length spacer molecules, were found to give an optimal rate of galactosylation when 10 spacer atoms were used (Ats et al., 1994). The lipophilic linker may have also increased the P 1,4-GalTase activity to a certain extent due to a previously observed hydrophobic effect (Geren, et al., 1976; Portner, et al., 1996). Another study reported that ovalbumin and the acceptor, benzyl-p-glucosamide, resulted in non-competitive inhibition of the P1,4-GalTase activity (Polentz et al., 1995). These results may be explained as two active sites being present on pl,4-GalTase, but equally pl,4-GalTase may only have one active site per molecule yet exist as a dimer (Malissard, et al., 1996). The N- and C- terminal pl,4-GalTase mutants produced by Boeggeman et al. (1995) show that the N-terminal fragment binds GlcNAc much tighter than the C-terminal fragment, but it is

unclear from this data if there are two GlcNAc binding sites on pi,4-GalTase (Dr. Pradman Qasba, National Institute of Health, Maryland, USA, personal communication).

In all the cloned pl,4-GalTases there are seven conserved Cys residues (Figure 1.3). It has been predicted that there is a disulphide bond Cys^^^-Cys^^S human p 1,4- GalTase that is essential for catalytic activity (Wang, et al., 1994). A similar observation has been made with the equivalent (Cys^^'^-Cys^^^) bond in bovine pl,4- GalTase (Boeggeman, et al., 1995). Interestingly, recombinant inactive P 1,4-GalTase mutants lacking the Cys^^^ ^^[\\ §^11 bind their substrates, albeit with a reduced affinity for the UDF compared to the catalytically active enzyme (Boeggeman, et al., 1995). Mutation of the carboxyl end Cys^^O iq ^ serine yielded a catalytically active pi,4-

GalTase with a much higher Km for UDP-galactose, further Cys^^O the only cysteine residue that reacts with sulphydryl reagents (Wang, et al., 1994).

The limited degree of homology between the primary sequences of the glycosyltransferases has been restricted to small areas of functional importance. C o m p ariso n of a sh o rt p ep tid e in pi ,4-GalTase, Cys3^0(X)5. 6(Lys/Arg)AspLysLysAsn(Asp/Glu), with that found in with mouse UDP- Gal:Galpl,4GlcNAc-R al,3-GalTase, Cys(X)5.6GlnAspLysLysHisAsp, has led to the suggestion that this region may form the UDP-galactose binding site (Joziasse et al., 1989; Yadav and Brew, 1990), a binding function both these transferases share. Yadav et al (1990) also found some correlation between these two transferases at sites further upstream of the hexapeptide that were involved in UDP-galactose binding. This extended region was very close to the hydrophobic pocket identified as the likely UDP- galactose binding site (Zu, et al., 1995). Comparisons of functional domains have been made between pl,4-GalTase and the glycosyltransferase UDP-GlcNAc: GlcNAcP-R P1,4-N-acetylglucosaminyltransferase from Lymnaea stagnalis which both transfer sugars onto GlcNAc residues in a pl,4 linkage (Bakker et al., 1994). High conservation of amino acids between these two transferases have been found in the area surrounding the hydrophobic domain (Bakker, et al., 1994), previously identified as involved in the catalytic site (Aoki, et al., 1990; Zu, et al., 1995). A number of the cysteines were also conserved between the two enzymes (Bakker, et al., 1994). Other glycosyltransferases, distinct from pl,4-GalTase but which share limited sequence homology and/ or function, have been identified (Bakker, et al., 1994; Neeleman and van den Eijnden, 1996). This opens up the possibility for the existence of a GalTase gene family. It is interesting to note that pl,4-GalTase can accept a variety of donor substrates, including UDP-Glc and UDP-GalNAc but not UDP-GlcNAc, yet at less than 0.4% of the rate for UDP-Gal (Palcic and Hindsgaul, 1991). Moreover, it has recently been demonstrated

that a-lactalbumin could induce pl,4-GalTase to efficiently use UDP-GalNAc in its transfer to GlcNAc, with the transfer of GalNAc only occurring to GlcNAc and not to Glc (Do, et al., 1995).

Divalent cations are an essential requirement for the pi,4-GalTase reaction, with Mn^+ as the most efficient (Navaratnam et al., 1986). It has been shown that there are two binding sites for the cations, a high affinity site I requiring micromolar concentrations of Mn^+ (apparent Km of approximately 10-20 |LiM) and a low affinity site II requiring millimolar concentrations of Mn^+ (apparent Km of approximately 800 |iM). pl,4- GalTase has approximately 70% full activity when site I is occupied and site II is empty, but is not active with only site II occupied (Kuhn et al., 1991). However, these requirements probably exceed the likely physiological concentration of Mn^+ by 2-4 orders of magnitude. When pl,4-GalTase was assayed within sealed Golgi membrane vesicles, apparent Km values towards Mn^+ of 0.1-0.2 jxM were observed with either added glucose or endogenous acid-precipitable acceptor (Kuhn, et al., 1991). The use of chelators in these vesicles permeabilized with A23187 almost abolished lactose synthetase activity, and was only adequately restored by Mn^+. Lysis of the vesicles caused the apparent Km at site I to increase to 10 |xM. Thus it appears that site I is probably occupied by Mn^+ in vivo (Kuhn, et al., 1991). Site II on the other hand can be stimulated by basic proteins such as histone and clupeine, as well as by small organic cations such as spermidine and spermine (Navaratnam, et al., 1986; Navaratnam et al.,

1988). Treatment of bovine lymphocytes and the epithelium of mouse small intestine with inhibitors of polyamine biosynthesis resulted in swelling of the Golgi apparatus and a decrease in pi,4-GalTase activity (Sakamaki et al., 1989). Addition of spermine or spermidine, restored the Golgi structure and also stimulated pl,4-GalTase activity (Sakamaki, et al., 1989). Whilst activation of pl,4-GalTase is achieved at both sites I and II, only site II stabilizes pi,4-GalTase (Kuhn et al., 1992).