The C. elegans intestine is derived from the embryonic blastomere E. The paradigm for E specification is that maternal SKN-1- and POP-1-dependent input cause activation of the E-specific gut specification factors
end-1 and
end-3. After reaching a threshold of expression, these activate
elt-2, which maintains its expression by autoregulation and drives the commitment to gut differentiation. Prior work suggests that gut specification may not be all-or-none, and that some of the descendants of E are capable of adopting a gut fate independently of others. We have created strains in which E is specified by single-copy transgene forms of
end-1 and/or
end-3 that are mutated for binding sites for the MED-1,2 GATA factors, themselves direct targets of SKN-1. In such strains, different embryos make variable numbers of apparently normal-sized gut cells, suggesting that specification has become subject to stochastic variations among embryos, and that commitment to a gut fate can occur later in the E lineage. Counting of embryonic
elt-2 transcripts by single-molecule FISH suggests that activation of
elt-2 is more graded in these strains, as opposed to an all-or-none mode as was previously reported for SKN-1-depleted embryos (Raj et al., 2010). As these effects are confined to the E lineage due to the nature of the strains constructed, we are able to evaluate adults derived from embryos in which functional guts were made. We find that such adults store lipids at significantly higher levels and display other variable pleiotropic phenotypes, suggestive of primary defects in gut function. Together, these results build a picture in which specification of gut is not an all-or-none event, and that in animals that do make an intestine, the endoderm differentiation network is not fully self-correcting for partially compromised specification. We will present these results, including the outcome of an attempt to computationally model the stochasticity of
elt-2 activation.