We are interested in identifying molecules involved in epithelial invagination and are using the formation of the C. elegans vulva as a model system. We have previously described the isolation and characterization of mutations that result in a defective invagination of the vulval epithelium (1). These mutations identify eight genes,
sqv-1 to 7 (squashed vulva) and
spe-2. The strongest alleles of each gene do not affect the vulval lineage or detachment from the cuticle but do result in an indistinct separation between the anterior and posterior halves of the invagination. Mutant adult animals are often egg-laying defective but can lay some eggs, indicating that their vulvas are at least partly functional. We have previously described the cloning of
sqv-3 and
spe-2 (1). The predicted SQV-3 protein is similar to members of a recently defined (2) family of glycosyltransferases that, so far, includes mammalian
beta-1,4 galactosyltransferase, snail
beta-1,4 N-acetylglucosaminyltransferase, several predicted human proteins of unknown function, and a C. elegans predicted protein, W02B12.11. In general, glycosyltransferases are Golgi enzymes that add specific sugars to the sugar side-chains of proteins in the secretory pathway. The predicted SPE-2 protein is similar only to a human protein of unknown function predicted from an I.M.A.G.E. Consortium expressed sequence. We have now also cloned
sqv-7, which lies on the cosmid C52E12. The SQV-7 protein, as predicted by Genefinder from data from the genome sequencing project, is similar to the LPG2 protein from the protozoan parasite Leishmania donovani (3).
lpg2- mutants are defective in the assembly of a complex cell-surface glycolipid, the lipophosphoglycan (LPG), that is required for a number of stages in the Leishmania life cycle, including infection of host macrophages. The LPG from
lpg2- mutants is missing repeating units of galactose(beta1-4)mannose(alpha1-PO4-6), as is a glycoprotein, sAP, but the glycosylation state of glycolipids and glycoproteins that do not contain this repeat is unaffected. This observation suggests that LPG2 is not required generally for glycosylation or Golgi function, but only for the formation of the specific linkages in the affected repeat. Since
lpg2- mutant extracts can still catalyze the assembly of the repeats when appropriate substrates are added exogenously, the LPG2 protein is probably not itself a glycosyltransferase but may be required for proper transport of the appropriate glycosyltransferase or its substrate into the Golgi (3). That the LPG2 protein is highly hydrophobic, as is SQV-7, is consistent with being a transmembrane transporter. SQV-7 is also highly similar to a human protein of unknown function predicted from an I.M.A.G.E. Consortium expressed sequence. The homologies of SQV-3 and SQV-7 suggest that the Sqv vulval phenotype can be caused by a defect in glycosylation. Because the screen for sqv genes is near saturation and has not hit at least six predicted homologs of glycosylation enzymes in the C. elegans genome, single mutations in these six glycosylation enzymes likely do not affect vulval invagination or else cause earlier severe phenotypes. Thus the Sqv phenotype is probably not simply caused by a susceptibility of vulval cells to a general glycosylation or Golgi defect. Instead, the lack of a specific sugar linkage and therefore perhaps a specific glycosylated protein containing that linkage may be required for wild- type vulval invagination. One possibility is that one of the uncloned sqv genes may encode the relevant protein substrate of SQV-3. 1. Worm Breeder's Gazette 14(1): 36 (1995). 2. Bakker et al. (1994). J. Biol. Chem. 269: 30326. 3. Descoteaux et al. (1995). Science 269: 1869.