SummaryWe have visualized the relationship between the endoplasmic reticulum (ER) and Golgi in leaf cells of Nicotiana clevelandii by expression of two Golgi proteins fused to green fluorescent protein (GFP). A fusion of the transmembrane domain (signal anchor sequence) of a rat sialyl transferase to GFP was targeted to the Golgi stacks. A second construct that expressed the Arabidopsis H/KDEL receptor homologue aERD2, fused to GFP, was targeted to both the Golgi apparatus and ER, allowing the relationship between these two organelles to be studied in living cells for the first time. The Golgi stacks were shown to move rapidly and extensively along the polygonal cortical ER network of leaf epidermal cells, without departing from the ER tubules. Co-localization of F-actin in the GFPexpressing cells revealed an underlying actin cytoskeleton that matched precisely the architecture of the ER network, while treatment of cells with the inhibitors cytochalasin D and N-ethylmaleimide revealed the dependency of Golgi movement on actin cables. These observations suggest that the leaf Golgi complex functions as a motile system of actin-directed stacks whose function is to pick up products from a relatively stationary ER system. Also, we demonstrate for the first time in vivo brefeldin Ainduced retrograde transport of Golgi membrane protein to the ER.

Homogenates of rabbit stomach mucosa were examined for enzymes catalysing the transfer of D-galactose from UDP-D-galactose to various low-molecular-weight acceptors of known structure. Treatment of the products with a and P-D-galactosidases revealed that D-galactose was transferred in both a and p-anomeric linkages. Competition experiments with the two Substrates indicated that the transfer of D-galactose was catalysed in each case by a different a-3-~-galactosyltransferase. Differences were also observed in the solubility properties of the enzymes: the a-3-~-galactosyltransferase using acceptor substrates with unsubstituted P-D-gdaCtOSyl residues was more readily soluble both in the presence and absence of detergents than the transferase using P-D-galactosyl residues substituted at carbon-2 with L-fucose. These findings demonstrate that rabbit stomach mucosa has two distinct a-3-~-galactosyltransferases : one, which is more tightly membrane-bound, resembles the human B-gene-specified transferase in its acceptor specificity, and the second, which is a more soluble enzyme, transfers D-galactose to the same positional linkage in unsubstituted P-D-galactosyl residues.

A monoclonal antibody was made after immunization of mice with the 1-32 amino terminal peptide of the alpha subunit of 32K human ovarian inhibin. The IgG2a mouse antibody reacted 6 times better with bovine 1-32 peptide than it did with 32K bovine inhibin. By contrast sheep polyclonal antibodies made by a similar method had a 29 fold bias in reactivity towards the immunizing peptide. Relative to homologous 1-32 peptide standards, the monoclonal antibody measured apparently higher amounts of immunoreactive material(s) in human (13.5 fold) and bovine (27 fold) follicular fluids than did the polyclonal anti 1-32 peptide antibodies. Immunochemical studies revealed that the epitope recognized by the monoclonal antibody was different from the major epitope recognized by the polyclonal antibodies. The monoclonal antibody reacted much better with human inhibin 1-32 sequences than with bovine (73 fold) or porcine (23 fold). Although the 32K form of human inhibin has not yet been purified, it can be inferred that the monoclonal antibody would be able to detect as little as 2 ng/ml of 32K human inhibin in competitive radioimmunoassays. The antibody must also react with some of the multiple molecular forms of inhibin found in human follicular fluids, and it was shown to function well in the quantitative immunoaffinity extraction of inhibin-like immunoreactivity from follicular fluid. It seems likely that this monoclonal antibody will prove a useful tool for research on human inhibin.

The alpha-2-L-fucosyltransferase in human plasma has been freed from alpha-3-L-fucosyltransferase activity and purified approximately 200,000-fold by a series of steps involving ammonium sulphate precipitation, hydrophobic chromatography on Phenyl Sepharose 4B and affinity chromatography first on GDP-adipate-Sepharose and then on GDP-hexanolamine-Sepharose. The purified alpha-2-L-fucosyltransferase had a M(r) on gel filtration HPLC of 158,000 and showed optimal activity in the pH range 6.5-7.0. The enzyme transferred fucose equally well to Type 1 (Gal beta 1-3GlcNAc) and Type 2 (Gal beta 1-4GlcNAc) substrates but Type 3 (Gal beta 1-3GalNAc) structures were less efficient acceptors. Competition experiments indicated that a single enzyme species in the purified preparation was responsible for reactivity with the Type 1 and Type 2 structures. Thus the differences in conformation between the Type 1 and Type 2 disaccharides do not appear to influence the capacities of their terminal non-reducing beta-D-galactosyl residues to function as acceptor substrates for the alpha-2-L-fucosyltransferase expressed by the blood group H gene in haemopoietic tissue.