Bridge over troubled waters

Spinal cord injury interrupts connections between the brain and spinal cord, rather than producing large-scale damage. Reconnecting severed axonswith their prior targets is a primary objective of spinal cord repair. Despite progress, this goal will probably not be attained soonbecausemanyproblemsremain to be solved.Wediscuss an alternative for promoting motor function after spinal damage by bridging the injury. We highlight a novel spinal injury bridge that we have developed to reconnect spinal motor circuits below the injury with thebrain. A spinalnerve thatexits above the injury is disconnected and inserted into the cord caudal to injury. Motor axons in the inserted nerve regenerate into the cord and synapse on neurons producing a novel circuit to bypass the injury. NeuroReport 15:2691^2694 c 2004 Lippincott Williams & Wilkins.


THE PROBLEM OF CNS AXON REGENERATION
Spinal cord injury (SCI) interrupts connections between the brain and spinal cord that transmit motor control signals and somatic sensory messages. There are over 2.5 million people worldwide with paralysis produced by SCI and over 90 000 new cases of traumatic SCI each year. The most obvious way to repair the spinal cord after injury is to promote axon regeneration to reconnect the severed axons with their original targets. This is not so simple. The spinal injury site is a maelstrom of molecules preventing damaged axons from regenerating. Among these are myelin proteins such as Nogo and MAG that become exposed after injury and which impede neurite outgrowth [1][2][3][4]. Blocking these proteins, singly and in combination, can promote some axon growth after SCI, as does inhibiting intracellular signaling pathways in the neurons that are activated by these proteins [4][5][6]. Nevertheless, even under the most favorable circumstances the number of regenerating axons appears to be too few to mediate substantial motor or somatic sensory recovery. SCI also results in the formation of a scar that is an impenetrable barrier to those axons that manage to avoid the inhibitory proteins [7]. This scar begins to form within hours of the injury, so treatments to promote regeneration must begin shortly after the injury.
Devising strategies to overcome SCI rely on animal models and there is the concern that some, and possibly all of the benefits of regeneration-promoting treatments are due to axons spared by the injury [8]. Contusion models of spinal injury, for example, leave a substantial portion of the white matter intact because tissue damage occurs centrally and partial SCI, such as dorsal hemisection, spares all ventral pathways. Even untreated animals with these models generally show significant improvement after injury and various treatments improve the rate of recovery. However, with key motor, sensory, and monoaminergic pathways intact, one is left with a disturbingly incomplete understanding of the mechanisms of recovery.
Unfortunately, promoting axon regeneration alone is insufficient to restore function. In addition to overcoming the hurdle of axon growth blockade, connections between regenerating axons and spinal neurons must be specific and appropriate. For example, neuropathic pain might be exacerbated if aberrant contacts regenerate between glutamatergic motor pathways and dorsal horn pain circuits. Similarly, unrestricted outgrowth of serotonergic terminations in the spinal gray matter may depress important protective signaling mechanisms by spinothalamic neurons. Without proper connections, regeneration is not only useless, but maladaptive.

ALTERNATIVE STRATEGIES FOR PROMOTING RECOVERY
While reconnecting severed axons with their prior targets is a primary objective of spinal cord repair, this goal will probably not be attained for some time. Parallel strategies for promoting motor and sensory recovery after SCI are called for. SCI typically interrupts connections between the brain and spinal cord, rather than producing large-scale damage to the cord itself, and spinal circuits below the lesion remain intact. One potentially effective strategy is to promote the function of these circuits. For example, applying monoaminergic drugs to the cord might promote locomotor function by increasing the ability of spared somatic sensory and motor pathways to regulate motor circuits below the lesion [9,10]. Similarly, activity-based rehabilitation strategies can be remarkably effective in animals and in humans [11,12].
An entirely different strategy is to engineer novel connections that allow motor control signals and sensory messages to bypass the injury. This strategy also exploits the fact that circuits caudal to the injury remain intact. There are four key criteria that determine whether a bridge that reconnects spinal motor and sensory circuits with the brain will be effective. First, the bridge must have high bandwidth, i.e. it must be capable of communicating a sufficient amount of information. The more axons in the bridge that can transmit information, the greater the bandwidth. Second, transmitting motor control signals caudal to the lesion requires that axons in the bridge originate above the injury so that they retain connections with supraspinal motor control centers. Similarly, to transmit somatic sensory information rostral to the injury, sensory axons in the bridge must originate from below the injury. Third, both sensory and motor axons within the bridge must achieve a precise and reproducible pattern of connections when they regenerate synapses on spinal neurons beyond the lesion. Fourth, supraspinal sensory and motor centers must be able to adapt to, and take advantage of, the new connections afforded by the bridge.
Over 20 years ago Alberto Aguayo and colleagues grafted a short segment of a peripheral nerve between the medulla and spinal cord and observed growth of CNS axons into the graft [13]. To be effective as a means to bypass a SCI, axons must grow through the bridge, emerge at the distal insertion site, and then grow into the cord to form new connections. More recently, this kind of bridge was used to connect medullary respiratory neurons with the cervical spinal cord after an upper spinal hemisection [14].
The major limitation to using an isolated nerve segment to bridge a spinal injury is that there is no a priori reason why any one type of neuron is more or less likely than any other to extend an axon into the graft. Given the diversity of CNS neurons, it is unlikely that every neuron phenotype will respond to trophic factors in the nerve bridge. Consequently, the number of any single neuron type regenerating into the nerve bridge is apt to be small and not reproducible. Important also, supraspinal motor systems might not be able to adapt internal representations to the novel connections when the populations of regenerating axons in the bridge are diverse or when the number of constituents in each group is small.
Although bandwidth is likely to be low, as will the potential of the brain to adapt to the novel bridge circuitry, this approach has promise and its ability to bypass an SCI might be improved. For example, the application of neurotrophic factors or other agents could increase the number of regenerating axons that enter the bridge and might facilitate synapse formation. Aguayo and colleagues adapted the original procedure to reconnect the eye and brain after optic nerve damage [15]. The population of neuron types that regenerated in this case (retinal ganglion neuron axons) was homogeneous, thereby enhancing the likelihood that the CNS target neurons will respond to the sensory signals transmitted by the bridge [15].

MOTOR NERVE BRIDGE
Building on these earlier studies, we have devised a novel bridging approach to bypass a spinal injury that has significant advantages over its predecessors and which has been shown to activate and promote the function of intrinsic motor circuits below a spinal lesion [16]. Our approach (Fig. 1) is to disconnect a spinal nerve that exits the cord above the site of the injury and insert the cut end directly into the lumbar spinal cord caudal to the level of injury (gray line). Motor axons in the inserted nerve regenerate into the spinal cord and synapse on local neurons. We have found that the bridge axons regenerate into intact spinal cord as well as the cord caudal to SCI. The vascular supply to the nerve is maintained, which appears to be crucial for its early survival. We selected the T13 nerve to insert because it innervates low abdominal muscles that collectively receive innervation by many thoracic nerves. Cutting one nerve does not produce any apparent loss of function.
The T13 nerve has about 80 motor axons, establishing an upper limit for regenerating axons and thereby information flow. When we injected anterograde tracer into the inserted T13 nerve we found that T13 motor axons, which are stained by cholinergic markers, regenerated into the spinal cord L-S T 13

From brain
To leg and back muscles To abdominal muscles  ( Fig. 2) and that most were directed towards the lumbar motor nuclei and portions of the intermediate zone that contain motor interneurons. Combined retrograde labeling of sciatic motoneurons in the lumbar ventral horn and anterograde labeling of regenerating T13 axons revealed direct contacts, many of which were on the cell body (Fig. 3a). By contrast, sensory axons in the inserted nerve appear not to regenerate into the cord [16]. This is not surprising since the peripheral axon of the sensory neurons does not synapse on neurons. The ability of T13 motor axons to regenerate into the cord was initially surprising. However, motor axons regenerate readily after peripheral injury [17] and there is evidence that other peripheral axons are not necessarily constrained by the processes that inhibit regeneration of central axons after injury. In a seminal study, Silver and colleagues demonstrated that dorsal root ganglion neurons transplanted into the CNS of adult rats can regenerate long distances in the normal and even degenerating CNS [18,19]. T13 motor axons clearly grow into the cord where they are likely to encounter various inhibitory proteins. We propose three mechanisms to explain the growth. First, the cut end of the nerve is inserted into the gray matter, which is likely to be more permissive to axon growth than white matter because myelin growth inhibitory proteins are not as prevalent. Second, the insertion causes minimal damage. When Silver and co-workers [18] implanted DRG neurons within the dorsal columns, they took care not to damage local axons. Limiting damage limits exposure of myelin growth-inhibiting proteins. Moreover, motoneurons may be relatively insensitive to neurite growth inhibitory cues since they do not normally encounter such cues in the periphery. Third, nerve insertion may have modified the local environment to make it more favorable for regeneration, perhaps by causing the release of factors from dedifferentiating Schwann cells at the cut end of the nerve [17]. This is reminiscent of the success achieved using peripheral nerve tissue to promote regeneration [13].
The formation of synapses between regenerating thoracic motor axons and spinal neurons also is surprising, but understandable because many ventral horn motoneurons have a collateral branch that synapses on Renshaw cells and motoneurons [20]. Thus, they are genetically endowed with the capability to form synapses on CNS neurons. Since the regenerating axons in the nerve bridge are cholinergic [16], target spinal neurons must be cholinoceptive for effective and sustained synapse formation. This suggests that as the regenerating axons emerge from the nerve bridge they seek cholinoceptive targets. If this is so, it constrains where these axons grow. Indeed, we found that regenerating T13 motor axons do not grow randomly within the gray matter, but rather are directed primarily to the intermediate zone and ventral horn. Not surprisingly, there is a good correlation between the distribution of regenerating T13 axons and the locations of cholinoceptive spinal neurons [21].
The synapses between regenerating T13 motor axons and spinal neurons are functional. Electrical stimulation of the inserted T13 nerve evoked complex spinal potentials that are localized to portions of the intermediate zone and ventral horn where the bridge axons regenerate. More important, stimulation evoked muscle contractions in the leg. The synapses between the regenerating T13 axons and spinal neurons appear to be constitutively active and can promote motor recovery. The lesion we make [16] is a spinal hemisection between the 2nd and 3rd lumbar segments, which produces wasting and spasticity-like signs in hind limb muscles that are progressive, as in human spinal cord injury. Regenerating axons in the inserted nerve ameliorates both the spasticity and muscle wasting.

REGENERATING T13 MOTOR AXONS CAN MEDIATE LOCAL CONTROL
Bridge axons regenerate up to about 1 mm rostral and caudal to the level of insertion. Growth presumably ceases once target neurons are contacted. The limited rostrocaudal growth is an advantage because the motor recovery mediated by the bridge axons appears to be by local control of motor nuclei. Motoneurons that innervate a variety of proximal and distal hind limb muscles are located in neighboring nuclei in the ventral horn (Fig. 3, right). Even with limited rostrocaudal growth, regenerating axons are in proximity to diverse motoneuron populations. Extending this logic to nearby premotor interneurons, more diverse control could be exerted, also without extensive rostrocaudal spread.
The cell bodies of the T13 motoneurons are located rostral to the injury and because they retain their supraspinal connections would function as interneurons in transmitting control signals from the brain to the spinal cord caudal to the injury. This is a novel approach to bridging a SCI. In contrast to the approach developed by Aguayo and colleagues, which promotes axon outgrowth into the nerve conduit and around the damage, the inserted motor nerve approach creates a new neural circuit to route signals past the injury.

PROSPECTS FOR DEVELOPING BRIDGE CIRCUITS TO RESTORE PARTICULAR MOTOR AND SENSORY FUNCTIONS
Our findings are very encouraging because the number of regenerating motor axons in the inserted nerve is high, they target specific spinal motor circuits, and they retain their connections with the brain because they originate from above the level of injury. The axons bridging the SCI access relatively intact motor circuits [22], with reflex connections and segmental control, suggesting that important hind leg inter-joint coordination is preserved, which is necessary for purposeful movements. What remains to be determined in future experiments is whether supraspinal motor centers, such as the motor cortex or red nucleus, can transmit control signals to motoneurons in the bridge to circumvent the injury.
Another important future direction is to provide a bridge for transmission of sensory messages around the injury, which is not a function of the motor nerve bridge. Aguayo and colleagues success in reconnecting the damaged optic nerve provide encouragement that a nerve conduit approach might be used to route ascending primary afferent fibers around a SCI. Recently Tadié and colleagues [23] have used Aguayo's approach in rats to guide dorsal root axons from below a SCI into the dorsal column above the injury. Nine months later, they used retrograde tracing to show that many dorsal root ganglion neurons had regenerated into the cord rostral to the injury.
It is important that a bridge around the damaged spinal cord avoids the scar that forms at the injury site. Since most research on SCI focuses on acute treatments, to initiate axon growth and avoid glial scar formation, this leaves little hope for patients with chronic injury. Given that the circuitry below the lesion remains intact after injury, the motor nerve bridge should be able to restore some motor function to patients with chronic spinal injuries, a potential not offered by treatments that only work acutely. The motor nerve bridge approach is highly flexible. Thus, a thoracic nerve inserted into the caudal sacral cord could be used to promote bladder function, whereas a branch of the spinal accessory nerve could be inserted into the cervical cord to promote forelimb motor functions. We can also insert several nerves to increase bandwidth for information transmission and can insert nerves at different rostrocaudal levels to target separate populations of motor nuclei [16]. By combining these different bridging strategies, sufficient information could be transmitted around the injury to achieve purposeful movements and to restore the protective aspects of somatic sensation.