Optimally controlling development, in theory

Summary by Katherine Rogers: Pezzotta, A., and Briscoe, J. (2023). Optimal control of gene regulatory networks for morphogen-driven tissue patterning. Cell Systems 14, 940-952.e911. 10.1016/j.cels.2023.10.004.

Image credit: Midjourney

Embryos reliably generate tissues containing cells with specific fates distributed in specific patterns. For example, in the mouse neural tube (the structure that becomes the brain and spine), one group of cells expresses a transcription factor (TF) called Nkx2.2 and will give rise to a specific type of neuron, while a neighboring group expresses the TF Olig2 and will produce a different neuron type. This spatial patterning requires a signaling molecule, or morphogen, called Sonic Hedgehog (Shh). How do morphogens reproducibly organize distinct cell fate domains?

Morphogens like Shh are often distributed in gradients. The elegant “French flag” model posits how gradients might organize patterning. If cell fates are specified by different morphogen concentrations, a gradient across the tissue would lead to spatially organized regions of fates [1-3]. Another influential concept, Waddington’s landscape, imagines cell fate decisions as a ball (representing a cell) rolling down a landscape of bifurcating valleys representing fates—terrain defined by the morphogen [4]. However, these morphogen-centric concepts can’t explain how embryos deal with real-world challenges like noise: Minor perturbations in morphogen distribution or gene expression would have catastrophic patterning consequences. Yet, embryogenesis is remarkably robust to these and other perturbations.

To discover what is missing, Pezzotta and Briscoe [5] assessed patterning systems through the lens of “Marr’s level of analysis” [6]. This information-processing approach considers systems from three levels: 1) the problem solved by the system (tissue patterning), 2) the relationships between system components that give rise to the solution (signaling in -> fate decisions out), and 3) the “hardware” underlying the physical system (morphogens, genes, etc.). They focus on level 2, defining a mathematical formalism describing the rules governing a theoretically optimal patterning system.

The authors created equations describing interactions between known components of the Shh patterning system, including Shh and the TFs Nkx2.2 and Olig2. They posited that an optimal patterning system should be fast, robust, and use minimal energy—it should require the least amount of input from the lowest number of components. They found that optimal systems consider signaling dynamics (i.e., changes over time), and heavily rely on “gene regulatory networks” (GRNs), the transcriptional responses downstream of morphogens. The Shh GRN features important cross-inhibitory interactions: Olig2 and Nkx2.2 are both activated by Shh, and each TF inhibits the other. This contributes to the mutually exclusive Olig2 and Nkx2.2 expression domains needed to correctly specify neuronal fates. A second important feature is that the Shh GRN is a “closed loop system”, feeding back on Shh itself by activating signaling modulators. This low-energy strategy can buffer signaling fluctuations.

In optimized patterning systems, the GRN is not a passive cog in fate decisions. In contrast to the French flag and Waddington models, where morphogens dictate the landscape, here the GRN is an equal partner in the decision-making process, and supports developmental robustness. Future experiments should test these ideas by probing developing systems with “open loop” approaches such as optogenetic signaling manipulation [7], and should determine how these principles apply to other patterning contexts, such as BMP/Nodal-mediated axis patterning.    

1. Sharpe, J. (2019). Wolpert's French Flag: what's the problem? Development 146. 10.1242/dev.185967.

2. Wolpert, L. (1968). The French Flag Problem: A contribution to the discussion on pattern development and regulation. In Towards a Theoretical Biology, C.H. Waddington, ed. (Edinburgh University Press), pp. 125-133.

3. Wolpert, L. (1969). Positional information and the spatial pattern of cellular differentiation. Journal of Theoretical Biology 25, 1-47. 10.1016/s0022-5193(69)80016-0.

4. Waddington, C.H., and Kacser, H. (1957). The strategy of the genes: A discussion of some aspects of theoretical biology. London, UK: George Allen & Unwin.

5. Pezzotta, A., and Briscoe, J. (2023). Optimal control of gene regulatory networks for morphogen-driven tissue patterning. Cell Systems 14, 940-952.e911. 10.1016/j.cels.2023.10.004.

6. Willshaw, D.J., Dayan, P., and Morris, R.G. (2015). Memory, modelling and Marr: a commentary on Marr (1971) 'Simple memory: a theory of archicortex'. Philos Trans R Soc Lond B Biol Sci 370. 10.1098/rstb.2014.0383.

7. De Santis, R., Etoc, F., Rosado-Olivieri, E.A., and Brivanlou, A.H. (2021). Self-organization of human dorsal-ventral forebrain structures by light induced SHH. Nat Commun 12, 6768. 10.1038/s41467-021-26881-w.