Nitric oxide gas has now been found to act as a switch during developmental remodelling of axonal projections from neurons: high gas levels promote the degeneration of unwanted axons and low levels support subsequent regrowth.
To create fantastic bonsai trees, a bonsai master prunes unwanted branches and promotes the growth of new ones with careful timing. Similarly, neuronal projections called axons must undergo proper and timely pruning and regrowth in the brain to produce functional neuronal circuits1. Failure of this process has been associated with autism and schizophrenia2,3. Until now, the way in which neurons transition between degenerative and regenerative states has been mysterious, but, writing in Cell, Rabinovich et al.4 report that the switch is mediated by levels of nitric oxide (NO) gas.
The mushroom body (MB) is a brain region in the fruit fly Drosophila melanogaster that is involved in associative learning and memory. During early pupal development, when larvae undergo metamorphosis into flies, the distal branches of MB axons are eliminated and then regrow, adopting different conformations that better serve the adult lifestyle. As such, MB axons offer an excellent model system in which to untangle the mechanisms that underlie neuronal remodelling5. Research on this system has provided a good understanding of axon degeneration6,7,8, but axon regeneration and the mechanisms that control the transition between the two states have not been well studied.
The group that performed the current study previously showed that, in D. melanogaster, a nuclear receptor protein called UNF is essential for axon regrowth9. In mice, the equivalent protein forms a dimer with another nuclear receptor, REV-ERB (ref. 10). Rabinovich et al. found that the fruit-fly equivalent of REV-ERB, a protein called E75, is also essential for axon regrowth. It has been proposed11 that haem molecules bind to each of UNF and E75, and that haem also binds to NO gas. In addition, NO levels modulate the activity of E75 (ref. 12). The authors therefore investigated whether NO is involved in axon regrowth during MB remodelling.
Using MB neurons in culture, Rabinovich et al. reduced NO levels by inhibiting the activity of the enzyme that catalyses NO production, NO synthase (NOS), either chemically or by inhibiting transcription of the NOS gene. Both treatments promoted regrowth of MB axons. By testing the physical interaction between UNF and E75, the researchers found evidence that the proteins interacted when NO was depleted, but not under normal conditions. Thus, they suggest that UNF and E75 form dimers that promote axon regrowth, but can do so only when NO levels are low. Moreover, depleting NOS in vivo caused not only precocious regrowth but also defective pruning, demonstrating the need for high NO levels during the degenerative phase of remodelling.
Next, the authors showed that NO levels in MB neurons undergo dynamic change during normal remodelling, being high during pruning and low during regrowth. However, levels of NOS messenger RNA and NOS remained unchanged during the transition between states. How, then, is the level of NO controlled? The NOS DNA sequence generates several mRNA isoforms, and Rabinovich et al. found that expression of at least one of these, which encodes a truncated form of NOS, coincided with regrowth but not pruning. NOS proteins must bind together into dimers to act enzymatically, so the production of truncated NOS isoforms might limit the capacity of even full-length NOS proteins to form functional dimers, severely decreasing NO synthesis.
To test this, the authors overexpressed full-length NOS in mutant MB neurons that lacked all NOS isoforms. As predicted, axon regrowth was drastically delayed. By contrast, regrowth was normal when full-length NOS was overexpressed in healthy MB neurons expressing the truncated NOS isoform. Rabinovich et al. therefore concluded that expression of the truncated NOS isoform does disrupt the formation of functional NOS dimers, causing a rapid drop in NO levels. This change allows the formation of UNF–E75 dimers, which activate downstream signalling pathways to promote axon regrowth (Fig. 1). How expression of the short isoform is controlled over time remains unclear, and identification of the underlying regulatory mechanisms will be the key to deciphering this.
NO is known to regulate the synaptic connections between neurons13, changing their strength in a gradual, activity-dependent manner. This regulation primarily involves the classic NO signalling pathway, in which NO induces production of cyclic GMP molecules through activation of the enzyme soluble guanylate cyclase, leading to local changes in synaptic regions of the cell. By contrast, Rabinovich et al. describe a process in which NO exerts acute, switch-like regulation. This difference can be explained by the fact that the regrowth switch acts not through the classic pathway, but through UNF and E75 — transcriptional regulators that probably act in the nucleus to modulate the expression of many genes after dimerization. This is a role for NO that was previously unknown.
Pharmacological inhibition of NO–cGMP signalling in photoreceptor neurons of the pupal fly brain induces the formation of disorganized and overextended axons14. It is intriguing that, even in the pupal brain, NO has different roles in different neurons and acts through different downstream targets. Thus, rapid changes in NO levels might simultaneously activate several developmental programs according to cell type.
How high NO levels promote axon degeneration remains unclear. During metamorphosis, NO-mediated E75 inhibition activates another nuclear receptor, FTZ-F1 (ref. 12). During MB remodelling, FTZ-F1 mutant MB neurons show pruning defects7, raising the possibility that NO-mediated E75 inactivation is required for pruning. However, Rabinovich et al. found that E75 mutants also had modest pruning defects. Thus, in the pruning phase, NO signalling probably acts through a different pathway (Fig. 1). Many signalling molecules have essential roles in MB pruning7, including TGF-β, the steroid hormone ecdysone and FTZ-F1. Testing the interactions between these signals and NO at high NO levels would help to reveal how NO promotes axon degeneration.
Given that mammals also have versions of the UNF, E75 and NOS proteins, and that the first two act as a dimer whose formation is probably affected by NO levels, it is plausible that a similar, albeit slightly different, molecular mechanism is found in humans, perhaps functioning during developmental remodelling in the brain. A connection between neurological disorders and defective neurodevelopmental remodelling is now becoming evident1. As such, it is worth investigating whether the NO switch acts in species beyond fruit flies.