Variability in gene expression contributes to phenotypic heterogeneity even in isogenic populations. Here, we use the stereotyped development of the Caenorhabditis elegans Q neuroblast to probe endogenous mechanisms that control gene expression variability. Posterior migration of the left Q neuroblast (QL) depends on a canonical Wnt signaling pathway which functions cell-autonomously to activate the expression of
mab-5/Hox. To identify the mechanism that ensures robust transcriptional activation, we used single molecule Fluorescent In Situ Hybridization (smFISH) to quantify the transcript levels of
mab-5 and other genes of the Wnt pathway in QL in a series of wild type and Wnt signaling mutant strains. Interestingly, we found that mutants that perturb Wnt signaling frequently exhibited increased variability in
mab-5 expression. Unexpectedly, these mutants also perturbed the levels of Frizzled (
mig-1/Fz,
lin-17/Fz and
mom-5/Fz) expression in QL, indicative of feedback regulations within the Wnt pathway. Combining computational network inference with quantitative gene expression profiling, we deduced a most probable network topology consisting of interlocking positive and negative feedback loops targeting the Frizzled receptors and
mab-5. Interestingly, model analysis suggests that positive feedback may cooperate with negative feedback to reduce variability, while keeping the expression of the target gene (
mab-5) at elevated levels. A minimal model of this signaling network predicts changes in gene expression variability across various mutants. Our results challenge the conventional view of the Wnt signaling pathway as a feedforward cascade and implicate gene regulatory network as an effective mechanism to ensure developmental robustness. In an ongoing effort, we attempt to elucidate the biochemical nature of these genetic interactions.