RESUMEN
In humans and Saccharomyces cerevisiae the free glycosylphosphatidylinositol (GPI) lipid precursor contains several ethanolamine phosphate side chains, but these side chains had been found on the protein-bound GPI anchors only in humans, not yeast. Here we confirm that the ethanolamine phosphate side chain added by Mcd4p to the first mannose is a prerequisite for the addition of the third mannose to the GPI precursor lipid and demonstrate that, contrary to an earlier report, an ethanolamine phosphate can equally be found on the majority of yeast GPI protein anchors. Curiously, the stability of this substituent during preparation of anchors is much greater in gpi7Delta sec18 double mutants than in either single mutant or wild type cells, indicating that the lack of a substituent on the second mannose (caused by the deletion of GPI7) influences the stability of the one on the first mannose. The phosphodiester-linked substituent on the second mannose, probably a further ethanolamine phosphate, is added to GPI lipids by endoplasmic reticulum-derived microsomes in vitro but cannot be detected on GPI proteins of wild type cells and undergoes spontaneous hydrolysis in saline. Genetic manipulations to increase phosphatidylethanolamine levels in gpi7Delta cells by overexpression of PSD1 restore cell growth at 37 degrees C without restoring the addition of a substituent to Man2. The three putative ethanolamine-phosphate transferases Gpi13p, Gpi7p, and Mcd4p cannot replace each other even when overexpressed. Various models trying to explain how Gpi7p, a plasma membrane protein, directs the addition of ethanolamine phosphate to mannose 2 of the GPI core have been formulated and put to the test.