(620f) Modeling Dorsoventral Patterning of the Drosophila Embryo in Silico Reveals Critical Details Overlooked By Fluorescence Imaging Studies

For more than 60 years, biologists have been investigating the chemical basis for the establishment of positional information within biological tissues, known as morphogenesis. During morphogenesis, a protein, known as a “morphogen”, conveys spatial information by forming a gradient across a tissue through various mechanisms, including local production, diffusion, and global capture. Differential gene expression based on morphogen concentration thresholds initiates tissue-specific responses that lead to cell differentiation and the establishment of distinct body structures. However, recent work modeling gene expression from fluorescence data has raised the question of how tissues far from the morphogen source can be patterned in the face of a low signal-to-noise ratio, causing us to question whether or not measured fluorescence is in fact proportional to the activity of a protein. Here we explore this question from the standpoint of patterning of the dorsal-ventral (DV) axis of the early Drosophila embryo.

The transcription factor Dorsal (dl), a homologue of the mammalian protein NF-kB, controls gene expression along the DV axis of the developing embryo. dl is ubiquitously expressed and sequestered to the cytoplasm by the inhibitor protein Cactus (Cact), which prevents it from promoting or repressing its target genes. However, signaling through the Toll receptor along the ventral side of the embryo causes Cact degradation and nuclear uptake of unbound dl protein in a ventral-to-dorsal gradient. Once dl becomes localized to the nucleus, it can activate or repress its target genes, including Snail (sna), Short Gastrulation (sog), and Decapentaplegic (dpp).

Despite recent developments in imaging techniques that have allowed us to quantify the dynamics of nuclear dl in live embryos, difficulties in modeling gene expression have persisted in regions where the signal-to-noise ratio is low. We hypothesize that this is a result of a misinterpretation of the data gathered with fluorescence imaging studies, which cannot distinguish between the fluorescence of unbound nuclear dl, which regulates gene expression, and nuclear dl-Cact complex, which does not. Previous to this study, it was assumed that dl-Cact complex was found only in the cytoplasm and therefore did not contribute significantly to nuclear fluorescence measurements. However, qualitative disagreement between the more recent experimental data and the established model caused us to question this assumption.  By modeling nuclear dl dynamics in silico, we were able to separate the two components and model gene expression under the sole influence of unbound dl.

This distinction between active and inactive pools of dl has allowed us to investigate the role of noise in establishing the borders of gene expression seen in the embryo. In fact, we propose that noise may play a significant positive role in determining the slope of gene expression boundaries.  Furthermore, we suggest that similar modeling work is needed in other cases in which fluorescence data cannot distinguish between an active form and an inactive form of a biological molecule.