In a 4-cell Caenorhabditis elegans embryo, two sister blastomeres called ABa and ABp are born with equivalent developmental potential, but eventually produce distinct patterns of cell fate. The different fates of ABa and ABp are specified at least in part by inductive interactions with neighboring blastomeres. Previous studies indicate that, at the 4-cell stage, a signal from the posterior-most blastomere, P2, is required for ABp to produce at least one of its unique cell types. This P2/ABp interaction depends on
glp-1, a putative receptor for intercellular interactions. To investigate this early induction further, we isolated mutants in which ABp developed abnormally. We describe the effects of recessive mutations in
apx-1, a maternal gene that appears to be required for P2 to signal ABp. In embryos from mothers homozygous for mutations in
apx-1 (
apx-1 embryos), ABp fails to produce its characteristic cell types. Instead, ABp from
apx-1 embryos develops more like its sister ABa: it produces ABa-like pharyngeal cells and it recapitulates ABa-like cell lineages. Because mutations in
apx-1 affect the development of only the ABp blastomere, we suggest that the wild-type gene encodes a component of the P2/ABp signalling pathway. To explain the observation that ABp in
apx-1 embryos adopts an ABa-like fate, we propose a model in which the P2 signal is required to break the initial equivalence of ABa and ABp. We performed two independent tests of this model. First, we examined ABp development in
pie-1 mutant embryos, in which P2 adopts the identity of another blastomere. We find that, in
pie-1 embryos, APp fails to produce its characteristic cell types and instead adopts a fate similar to that of ABa. We conclude that the changed identity of P2 in
pie-1 embryos prevents the P2/ABp interaction. As a second test, we examined ABp development in wild-type embryos after physically removing P2. These operated embryos produce extra pharyngeal cells, consistent with out proposal that a signal from P2 breaks the initially equivalent developmental state of ABa and ABp. We discuss the possibility that
apx-1 acts as a ligand in this
glp-1-dependent signalling pathway.