When we learn science in school, we are often taught the “facts”: the Earth is round, photosynthesis produces oxygen, and DNA contains the genetic code of an organism. But where did these facts come from? In the process of understanding the natural world, scientists go through multiple rounds of observation, hypothesis generation, and experimentation—revising models depending on the results. When a model has sufficient evidence without arguments to refute it, the model becomes generally accepted as “fact”. However, even textbook models can be challenged in the face of new evidence.
A pair of recent papers by Dominici et al. and Varadarajan et al. challenge a model that has dominated neuroscience for the past two decades—demonstrating the scientific process in action. The question addressed in these papers relates to how neurons in the developing nervous system can make the right connections. A functional nervous system that allows us to respond to our environment depends on the formation of specific connections between cells: for example, a neuron responsible for moving leg muscles needs to connect to a leg, and not to the stomach. But how do these specific connections arise? That is, how can a developing neuron recognize its correct target, while avoiding erroneous intermediate targets?
Based on the past two decades of experimentation, scientists developed a model which explained the general strategy a neuron could use to find its correct targets. This classical model explained how specific neurons, called commissural neurons, can extend axons that cross from the left side of the spinal cord to the right, or vice versa. This process of midline crossing presents a challenge because the axon needs to first navigate toward the midline, and then move away from the midline to its final destination—as if the axon were making an intermediate stop on a GPS route. How can molecular cues allow commissural axons to cross the midline without getting stuck, or crossing over multiple times?
In the classical model, the midline has a structure called the floor plate, which produces two types of chemical signals: Netrin and Slit. Netrin and Slit then diffuse out from the floor plate to form long-range gradients that are recognized by commissural axons. Before crossing the midline, commissural axons only have a receptor for Netrin, which attracts the axons to the midline—like how you might follow your nose to the pie in the kitchen. Once at the midline, the axons are able to switch receptors to recognize Slit instead, which repels the axons away from the midline. Indeed, previous work showed that mice completely lacking Netrin or Slit have defective midline crossing, indicating that Netrin and Slit are required for this process.
This well-established model has made its way into introductory neuroscience textbooks and serves as the classic example for how axons find their correct targets. However, the papers by Dominici et al. from the University of Lyon and Varadarajan et al. from UCLA challenge this model. The authors make use of a genetic toolset that had not yet matured when the classical model was created—the ability to specifically delete a gene in a subpopulation of cells in mice. By genetically deleting Netrin only from floor plate cells, the authors show that commissural axons can still cross the midline (Dominici et al., Figure 2 and 3; Varadarajan et al., Figure 1). This result indicates that floor plate Netrin is not needed for midline crossing, conflicting with the classical model that a long-range Netrin gradient attracts the commissural axons to the floor plate.
How can we reconcile this result with the previous finding that mice completely lacking Netrin have defective midline crossing? By specifically labeling cells that make Netrin, the authors show that Netrin is also secreted from another set of cells, found in the ventricular zone (Dominici et al., Figure 1, Varadarajan et al. Figure 1). Interestingly, this ventricular zone source of Netrin was actually discovered along with floor plate Netrin back in 1996, but was essentially ignored until now (Serafini et al.). When Netrin was specifically deleted from ventricular zone cells, but not from floor plate cells, commissural axons failed to cross the midline (Dominici et al. Figure 4; Varadarajan et al. Figure 2). This observation suggests that ventricular zone Netrin is responsible for guiding axons to the midline, not floor plate Netrin as previously thought. However, ventricular zone Netrin would not generate a long-range gradient for attracting commissural axons, so Dominici et al. and Varadarajan et al. propose a model where ventricular zone Netrin acts as a short-range cue—commissural axons crawl between local Netrin sources, much like climbing the rungs of a ladder.
By showing that floor plate Netrin is not necessary for midline crossing, Dominici et al. and Varadarajan et al. prompt us to re-examine a classic model for how neurons find their correct targets. Specifically, these papers question the significance of “long-range” chemical signals that guide commissural axons. It is important to note that many aspects of the classic textbook model were not challenged—in the revised model, Netrin still attracts axons to the midline, and Slit still repels axons after they cross the midline. Furthermore, the new model does not explain why axons crawl in a particular direction on the Netrin ladder (“up” the ladder vs. “down” the ladder), which future experiments will need to address. Still, this paper highlights the evidence-based, iterative scientific process: old models are revised according to new observations made possible by new scientific tools, creating new models which must face their own test of time.
- Dominici, C. et al. Floor-plate-derived netrin-1 is dispensable for commissural axon guidance. Nature (2017). doi:10.1038/nature22331
- Varadarajan, S. G. et al. Netrin1 Produced by Neural Progenitors, Not Floor Plate Cells, Is Required for Axon Guidance in the Spinal Cord. Neuron 94, 790–799.e3 (2016).
- Serafini, T. et al. Netrin-1 Is Required for Commissural Axon Guidance in the Developing Vertebrate Nervous System. Cell 87, 1001–1014 (1996).