'Above C4, breathe no more.' This is the memory aid that reminds medical students that damage to the spinal cord above the fourth cervical vertebra (C4) — that is, the neck — can interrupt breathing. Injuries at the cervical level are the most common type of spinal-cord injury and account for more than half of all cases. Individuals who survive such injuries usually need ventilators to breathe, and so face a host of complications to their overall health and quality of life. A study by Alilain et al.1 on page 196 of this issue offers hope that we may one day know how to treat this problem, so that patients with spinal-cord injuries above C4 can breathe on their own.

Breathing rate, rhythm and depth are controlled automatically by specialized regions of the brainstem2 (Fig. 1a). The neurons in these regions send their axonal processes down the spinal cord to control the activity of other neurons in the phrenic motor nuclei (PMN) of the cervical spinal cord (C3–C6). The axons of the PMN neurons form the phrenic nerves, which, in turn, innervate the muscles of the diaphragm. Thus, contraction and relaxation of the diaphragm enable rhythmic breathing. When the spinal cord is injured above the C4 level, axons connecting the brainstem to the PMN are damaged, and breathing is disrupted. To make matters worse, axons in the adult spinal cord do not regenerate well, one of the main reasons being the inhibitory environment of the injured spinal cord3.

Figure 1: Experimental manipulation of the respiratory system.
figure 1

a, Basic anatomy of the intact respiratory system. BS, brainstem; SC, spinal cord; C, cervical level; PMN, phrenic motor nucleus; PN, phrenic nerve. b, After a partial spinal-cord injury at C2, innervation to the left PMN is interrupted and the diaphragm is paralysed on the left. c, Alilain et al.1 show that 12 weeks after peripheral-nerve grafting and injection of chondroitinase ABC (chABC), activity is restored in the paralysed side of the diaphragm.

Over the years, researchers have invoked many strategies to provide axons with a more supportive environment. These include either removing inhibitory molecules, such as chondroitin sulphate proteoglycans (CSPGs) in the extracellular matrix4, or grafting in a piece of peripheral nerve that could serve as a bridge for axonal growth5. Combinations of these approaches have yielded encouraging results. For example, after a cervical spinal-cord injury (SCI) in rats, applying a peripheral nerve graft, together with injection of the enzyme chondroitinase ABC (chABC) to degrade CSPGs, allows spinal-cord axons to regenerate through the graft, re-enter the spinal cord and form synaptic connections with neurons on the opposite side of the injury6.

Alilain et al.1 applied a similar treatment strategy to recover respiratory function in rats after SCI. The authors made a partial injury at the C2 level to paralyse the diaphragm on one side of the animals' body (Fig. 1b). They then removed a piece of the rats' tibial nerve and grafted one end of it in the injury site at C2 and the other end in a small slit at the C4 level — near the PMN. Finally, they injected chABC at both ends of the graft, as well as in the PMN area, to degrade CSPGs (Fig. 1b).

Twelve weeks after injury, the group receiving this treatment had the highest percentage of recovered animals and the best quality of recovery in respiratory function compared with controls. Specifically, in many animals the paralysed half of the diaphragm muscle recovered nearly normal rhythmic electrical activity. Moreover, neurons from breathing centres of the brainstem grew axons into the graft. To demonstrate that recovery was largely due to axons that had regenerated through the graft and not just the rewiring of circuits in the portions of the spinal cord that were uninjured, Alilain et al. cut the graft; this treatment abolished the regained respiratory function.

Intriguingly, when the graft was cut, the residual electrical activity of the paralysed half of the diaphragm muscle was significantly, albeit transiently, increased compared with residual activity after the initial SCI. This suggests that spinal-cord circuits were also rewired to some extent in this experimental setting: descending regenerating axons may have connected with different targets, and denervated neurons may also have found new synaptic partners before regenerating axons could reach them7. It is encouraging that, despite the havoc such rewiring could wreak, the system could still adapt to the changes — perhaps through a type of learning process — to restore proper firing patterns to the motor neurons innervating the diaphragm.

Such adaptability might be a special property of the respiratory circuit: it has long been known8 that there are latent connections in the respiratory system that can be activated after injury. Alternatively, it may be a common property of all spinal-cord circuits that can be unleashed by the degradation of CSPGs, as well as by extensive rehabilitation. At an axon's target site, such as the PMN, CSPGs are thought to stabilize circuits after development is complete9. Degradation of CSPGs after injury may therefore give circuits the flexibility they need to make new connections and adapt to changes. The built-in rehabilitation regimen that the paralysed diaphragm receives — by virtue of the fact that the animal must continue to breathe if it is to remain alive — probably also plays a part in shaping the new circuit. In support of this, chABC treatment together with physical rehabilitation promotes recovery of manual dexterity after cervical SCI in rats10.

Overall, Alilain and colleagues' results1 suggest that combinatorial strategies that promote both long-distance axon regeneration and local circuit reorganization may be universally useful for enabling functional recovery after SCI. Because not all axons of the central nervous system have the same ability to regrow into permissive grafts11,12, it may be necessary to use other methods to stimulate axon regeneration13. Future studies should investigate how best to facilitate integration of regenerated axons into local circuits and to harness the potential of anatomical plasticity to restore multiple functions after SCI.