III.A GENERAL CONSIDERATIONS
The hyper complexity of the CNS organization combined with local dynamic and local remodels explains why the human kind is endowed with the most hyper-sophisticated cognitive functions of the animal kingdom. However, as remarked Cajal, one striking feature of such a highly developed nervous system is that: “Once development was ended, the founts of growth and regeneration of the axons and dendrites dried up irrevocably. In adult centres the nerve paths are something fixed, ended, immutable. Everything may die, nothing may be regenerated”. In more prosaic terms, the other hallmark of the adult mammalian CNS is related to its extreme vulnerability to injuries and to recover from them. Definitely it is of common knowledge that injuries of all kind like traumatic brain injury, Stroke, cerebral ischemia, neurodegenerative diseases such as Alzheimer and Parkinson Diseases, Multiple Sclerosis (etc.) lead to more or less irreversible damages and a definitive alteration of CNS function. This statement remains such of veracity that when we meet people victim of a spinal cord injury (SCI), we perfectly know, that they are condemned to remain in a wheelchair for the entire rest of their life. This cold statement reflects our actual impotence to resolve definitely the nervous injury related problematic.
III.B THE SPROUTING VERSUS NERVOUS REGENERATION
Basically the weakness of superior vertebrates CNS to functionally recover is firstly attributable to the fact that, after birth, dead neurons are almost never substituted by new neurons. In mammalians the neurogenesis occurs mainly at the prenatal period. At the adulthood this process still exists but is localized in few seats (e.g. the subventricular zone) while in low vertebrates, like birds, the neurogenesis is persistent and rather widespread distributed in all CNS (Kaslin et al. 2008).
Consequently the massive neuronal death following injury results de facto in the degeneration of the axons and their synaptic contacts. This leads finally to an alteration or even more to the annihilation of the entire function (motor, language, memory etc.). Therefore, in absence of neuronal turn-over, the character more or less reversible of the injury is primarily determined by the nature and the duration of the injury (e.g. transient ischemia duration).
However it is also dependent on the endogenous mechanisms of functional recovery (in term axonal growth and synaptic contact replacements) generated by non-dying neurons. Indeed following an injury the nervous system is capable to trigger mechanisms of auto-repairing displaying thus a kind of flexibility. These processes are the axonal sprouting and the regeneration (figure 8).
Figure 8: Reactive axonal sprouting versus regeneration.
After an injury, the CNS can mobilize a number of responses. If neurons degenerate as represented in A, the response is that the axons of healthy neurons sprout and form new connections with the target of the dead neurons. A second possibility, represented in B is that the severed axon regenerate but this does not occur. The damaged cells can increase collateralization on available target cells, thus the circuit is rebuilt (Adapted from Siegel, 1999)
In fact most of the time, a substantial number of neurons survive inside and also in the vicinity of the lesion site. Some of them are spared and functional. Concerning this category of neurons, they can sprout axons and establish new synaptic contacts (reactive synaptogenesis process) with the primary target of the dead or lesioned neuron (Yiu & He 2006). Under some circumstances this mechanism which is also named “reactive collateral sprouting” appears to be relatively efficient to compensate the disappearance of synaptic contacts. For example it explains why the triggering of the Parkinson disease associated symptoms (that is caused by massive death of dopaminergic neurons of the Substantia Negra) occurs only once more than half of the neurons are dead.
Nevertheless the axonal sprouting possesses a very limited efficiency in case of severe lesion such as spinal cord injury (SCI) or for advanced phase of neurodegenerative disease since such a mechanism, indeed, involves small populations of neurons (Cafferty et al. 2008).
However, a majority of neuronal cells are lesioned and exhibit a complete axonal transection and thereby become functionally ineffective. Generally in these neurons the distal part of these sectioned axons undergoes Wallerian degeneration while the proximal part becomes atrophic and forms at its extremity a “dystrophic growth cone” (Figure 9) or, as described by Cajal, quiescent
“dystrophic endballs”.
Figure 9: Schematic representation of the sequence leading to the formation of a growth cone.
Injuries to CNS neurons lead often axonal transection and generate debris such as the myelin sheath that wrapped the sectioned axon. While the distal degenerates (Wallerian degeneration), the proximal part of the axon becomes atrophic and forms a dystrophic growth cone that is most of the time incapable to establish new synaptic contacts.
In mammals these axotomized neurons are most of the time incapable to generate de novo long distance axonal extensions in order to reestablish synaptic contact(s) with their primary synaptic target(s). In brief, adult mammalian CNS neurons do not spontaneously regenerate following injuries. This latter sentence matches the Cuvier’s (1769-1832) principle. In fact, adult low vertebrates such as fish and amphibians can regenerate part of their body including significant portions of their CNS (Ferretti et al. 2003, Kaslin et al. 2008). In contrast mammalian central neurons progressively lose this capacity along with the progression of the ontogenesis: they display an age-dependent capacity to regenerate or, in other words, they lose their intrinsic regenerative abilities.
Implicitly this also indicates that mature neurons are incapable to recapitulate the developmental-related mechanisms that govern and lead to the set-up of the axonal wiring and synaptogenesis.
This is the reason why the analysis and comprehension of the mechanisms responsible of the building of the nervous system is susceptible to provide supportive informations about the CNS regeneration problematic in view of potential therapeutics.
III.C THE AXONAL GUIDANCE AND THE GROWTH CONE
III.C.1 The growth cone
In fact one can primarily consider that in both analog “situations” (i.e. developmental versus lesional neurons) the finality is similar in terms of axonal motility and synaptogenesis and thereby that the basic mechanistic and factors employed by embryonic and mature neurons is a fortiori identical. Indeed in both contexts the common point is the formation of a “growth cone”.
The growth cone is the structure localized at the extremity of the developing/regenerating axons which is both responsible of the perception of the environment and which also guides the axonal extension to its final destination. It is thus of interest to consider the determinants that act on this key structure during development. The growth cones were initially observed in the 1890s by Cajal and envisioned that it was “....endowed with exquisite chemical sensitivity, rapid ameboid movements and a certain motive force thanks to which it is possible to proceed forward and overcome obstacles met in its way…until it reaches its destination. To summarize, Cajal by simply observing static images, knew that this “club shaped structure”, has a sensor and motor function that consist of exploring the environment and guiding the axon to its final destination where it establishes synaptic connections.
Description
The growth cone is localized at the end of the developing neuron (figure 10). Their basic structure is composed of two main regions: a central core that is rich in bundled microtubules and a periphery that is rich in F-actin (F for fibrillar) filaments that derived from polymerization of monomeric G-actin (G for globular) (Dent & Gertler 2003, Kalil & Dent 2005, Huber et al. 2003). In the peripheral region we distinguish two structures: the lamellipodia localized at the proximal part of the growth cone and the filopodia, a “finger-like” shaped structure localized at tip of the growth cone. In lamellipodia, F-actin is organized in cross-linked network forming a meshwork (or “veil-like” structure). This organization is gradually replaced toward the leading edge of the growth cone by F-actin organized in bundles.
Figure 10: The growth cone.
Left panel: the growth cone is composed a central part (C) and a peripheral part (P). The peripheral part is divided in two regions the lamellipodia and the filopodia, at the leading edge of the growth cone. Right panel: the core is rich in microtubules (anti-tubulin) while the periphery is rich in actin filament.(from (Gallo & Letourneau 2004).
The chemoaffinity hypothesis
In 1963 Sperry (Nobel prize in 1981) proposed the “chemoaffinity hypothesis” for the axonal guidance. Growth cone guidance is ruled-out by the presence of “tags” on the growth cone that respond to gradients of chemical cues that guide it, by attraction/repulsion mechanisms, to its final destination (Sperry 1963). More precisely the axon navigation is directly determined by extracellular guidance cues that bind the growth cone surface receptors which, via specific intracellular signaling pathways act ultimately on the cytoskeleton dynamic, thereby determining the axonal direction and elongation (Kalil & Dent 2005, Huber et al. 2003, Tessier-Lavigne &
Goodman 1996, Gallo & Letourneau 2004).
These chemical cues guide the neurons through four mechanisms (Figure 11): by chemoattraction (netrin for example), by chemorepulsion (secreted semaphorins) that supposes a long-range action of these cues or by contact repulsion (Semaphorin, tenascin) and by contact attraction involving thus a short-range action of the target cell (Chilton 2006, Tessier-Lavigne & Goodman 1996). The contact attraction mechanisms requires that the substrate must be adhesive and permissive as well (i.e. some molecules have adhesive property without promoting axon growth). The molecules that constitute this attractive substrate are in general the members of the extracellular matrix molecules (ECM) like for example the laminin, and the CAM (Cell Adhesion Molecules) for cell-cell interaction.
Figure11: Guidance force of the axonal navigation.
The guidance cues can be classified in two parts: long range cues: by chemical or release molecular factors that can be attractive like the Netrins or repulsive like the semaphorins or short range cues: acting by contact that also can be attractive like the extracellular matrix molecules (ECM) like laminin and cell-adhesion molecule (CAMs) or and cadherin or repulsive like the ephrin family. (Adapted from. (Huber et al. 2003))
The growth cone cytoskeleton dynamics
The underlying mechanisms of axonal growth are known to involve precise and intricate cytoskeletal modifications of the actin dynamics within the growth cone (Zhou & Cohan 2004, Dent
& Gertler 2003, Goldberg & Burmeister 1986). Nonetheless, increasing evidences show that the cytoskeleton dynamics involves modifications of the microtubules in the core of the growth cone as well (Zhou & Cohan 2004, Kalil & Dent 2005).
These cytoskeleton dynamics have primarily described in the peripheral region i.e. in term of actin cytoskeleton dynamics (figure 12). Indeed it appears that filopodia and lamellipodia growth is accounted by actin filament polymerization at the leading edge of the growth cone and actin depolymerisation in the proximal part of the growth cone (Dent & Gertler 2003, Luo 2002, Kalil &
Dent 2005, Forscher & Smith 1988). Forscher and Smith, by coupling high-resolution microscopy and an inhibitor of polymerization (cytochalasine), have emphasized that filopodia subsequently
disappear. They demonstrated therefore that the polymerization process is accompanied by a retrograde flow of F-actin from the leading edge to the proximal part of the growth cone where depolymerisation process takes place (Forscher & Smith 1988). In other words, the state of the filopodia extension is the net effect between the rate of polymerization and the rate of the actin retrograde flow (i.e. attractive cues would promote actin polymerization and in the same time prevent retrograde actin flow and reversely concerning the action of repellents). The rate of retrograde actin flow is determined by acto-myosin II contractions whose activity is controlled by phosphorylation (Brown & Bridgman 2003). Thus phosphorylation of myosin increases the speed of retrograde flow and leads to the retraction of the filopodia.
.
Figure 12 : Actin dynamics in growth cone.
The fiipodia and lamellipodia growth is accounted by the polymerization of actin, at the leading edge of the growth cone (+) and depolymerisation in the proximal part (-) of the growth cone. These chemical processes are accompanied by a retrograde flow of F-actin. Hence, a repulsive cue increases both the F-actin polymerization as well as the speed of F-actin retrograde flow leading to the retraction of the filopodia. (Adapted from (Huber et al. 2003)
The action of the Rho-GTPases on the growth cone cytoskeleton
How guidance cues ultimately affect cytoskeleton organization? They are thought to likely act on small GTPases that are well-known to regulate actin organization in non-neuronal cells (Hall 1998).
Thus the actual paradigm is that Rho GTPases play an instructive role (activated directly by the cue) in that their activity on cytoskeleton actin dynamics is directly regulated by receptors of the
guidance cues (Dickson 2001). F-Actin filament assembly and disassembly (through the mechanisms of polymerization/depolymerisation), and retrograde flow have recently been shown to be directly dependent of the Rho GTPases activity and their downstream effectors (Menager &
Kaibuchi 2003, Dickson 2001). These proteins belong to the superfamily of small GTPases ras (Gallo & Letourneau 2004, Luo 2002, Menager & Kaibuchi 2003, Peck et al. 2002). Like the G-protein ras, Rho G-proteins cycle between two distinct forms (Figure 13). They are active (i.e. able to act on effectors) when bound to the GTP and inactive when they are bound to the GDP, after hydrolysis of the GTP (Ramakers 2002). Their activity is regulated by guanine nucleotide exchange factors (GEF) and GTPase activating proteins (GAPs) and GDP dissociation inhibitor (GDI). GEF facilitate the exchange of GDP to GTP activating thus the Rho GTPase signaling, whereas GAPs and GDI activate their endogenous GTPase activity that is of low level (Peck et al. 2002) turning-off thus the Rho-linked signaling.
Figure 13: The RhoGTPase cycle.
(from (Luo 2002)
The most studied members of this family are RhoA, Rac1 and Cdc42. Their role is not identical and even opposite. Hence it is the asymmetry of their activity within growth cone that determines the axonal growth (Dickson 2001).
The different advances concerning the underlying mechanisms leading to the synaptogenesis are logically helpful. Indeed they permit to identify some basic molecular and mechanistic components that are highly susceptible to be identical given the fact that axonal lesion leads also to the formation of a growth cone. In consequence, by proceeding to a differential strategy between these two contexts (embryonic versus mature CNS) one could then figure-out what are the extrinsic/intrinsic factors that account for the decline in the axonal growth ability.
For example one strategy to identify molecules that may participate in regeneration is protein electrophoresis from nervous tissues during regeneration (regenerating PNS neurons) followed by a comparison with the same tissue under control condition. Using this differential approach, several rapidly transported growth associated proteins have been identified. The first and best known of them is the growth-associated protein of 43kDa (GAP-43).
III.C.2 The Growth associated protein-43
In 1981, Skene and Willard reported that a growth cone protein is expressed in all neurons during the development and then downregulated. They also reported the systematic reappearance of this protein in association with successful axonal regeneration of CNS neurons and absent in nerve that do not regenerate (Bulsara et al. 2002, Benowitz & Routtenberg 1997, Skene 1989, Skene &
Willard 1981). This small protein, acidic membrane protein was the named growth associated protein of 43 kDa (GAP-43). The precise function of GAP-43 remains unknown but this small protein has been revealed to be associated with the growth cone membrane, to be phosphorylated by PKC and binds to calmodulin (Benowitz & Routtenberg 1997).
GAP-43 is robustly re-expressed after peripheral nerve injury (Skene & Willard 1981) and in contrast CNS lesions activate only a fraction of the injured neuron (Schaden et al. 1994). Later is has been shown that the overexpression of this protein induces sprouting in vivo (Aigner et al.
1995) whereas GAP-43 deficient mice show defects in axonal pathfinding (Strittmatter et al. 1995).
Very interestingly, Bomze et al reported that the overexpression of GAP-43 alone induces axonal sprouting but do not trigger regeneration in an in vitro model spinal cord transection. A similar result was found with the overexpression of another growth cone protein, the cytoskeleton-associated protein of 23kDa (CAP-23). However the combined overexpression of GAP-43 and CAP-23 supported the axonal regeneration in a large population of DRG neurons as well as in vivo (Bomze et al. 2001).
In regard to this chapter relative to GAP-43 the age-dependent regenerative profile comfort more again an individual to think that the inability to regenerate is mainly due to the loss of components that are purely intrinsic to the neurons.
III.D THE GLIAL INHIBITION AND THE INTRINSINC GROWTH STATUS
III.D.1 The glial inhibition of the central nervous system regeneration
The “intrinsic hypothesis” is challenged by the fact that PNS neurons exhibit throughout their life a strong regenerative ability (Schwab & Thoenen 1985). Furthermore, neurons such as those in dorsal root ganglia (DRG) have axons in both the CNS and PNS, but can only regenerate their peripheral processes. These observations strongly suggest that it is in fact CNS factor(s) of an external nature which inhibit(s) the nerve regeneration (Qiu et al. 2000, Yiu & He 2006). In 1981 this new paradigm (even if Cajal, here again, had already suggested this in 1905) was empirically demonstrated by David and Aguayo who emphasized that the axon growth of CNS lesioned neurons is increased if they were provided with a peripheral environment which is radically different (e.g.
Schwann cells instead of oligodendrocytes) from the CNS (David & Aguayo 1981). As a consequence concept of “glial inhibition” was born.
From now on it appears that the post-natal development of CNS glial cells and their degeneration (i.e. the myelinic debris) coupled to their reactions consecutively to a given insult are responsible for the progressive decline of axon regrowth. In other words, the neurons do not lack an intrinsic ability to regenerate. This first decisive step will be followed in the early 1990’s by the discovery of the first inhibitory molecules, their receptors as well as their signaling.
The Oligodendrocytes/myelinic inhibitor, Nogo
In the late 1980’s Martin Schwab and colleagues, from the University of Zurich, isolated a monoclonal antibody that neutralize the action of a fraction of myelin, termed IN-1, which was previously shown to inhibit neurite extension in vitro (Schnell & Schwab 1990, Caroni & Schwab 1988). Moreover when the antibody against IN-1 was injected into adult rat spinal cord injury, about 5% of the severed axons regenerated across the injured tissue and the animals showed striking functional recovery of their motor functions (Bregman et al. 1995). These spectacular data obtained with this antibody became definitely revolutionary when they were also confirmed in adult monkeys that recovered a complete manual dexterity after cervical lesion (Freund et al. 2006).
Naturally these results obtained in a pre-clinical test (carried-out by Novartis and Martin Schwab) have raised high hopes in this strategy of therapy applied to CNS injuries in human. Consequently a phase I clinical trial of intrathecal infusion of anti-IN1 is being carried-out in Europe and one other will begin in North America (Ruff et al. 2008).
Later the antigen for the IN-1 antibody was cloned and was named Nogo (in reference to the
McMahon 2006, Yiu & He 2006, Filbin 2003). Nogo has been shown to belong to the reticulon family and to exist in three distinct forms (Nogo-A,-B and -C) that are generated by alternative splicing (Figure 14). All three Nogo have a common carboxyl (C) terminus. Among these, Nogo-A is the best characterized due to its high expression in oligodendrocytes and myelin (Huber et al.
2002). Structure-function analysis revealed that two inhibitory domains are present in NogoA: a unique amino (N) terminus (amino Nogo) and a 66 amino acids loop (Nogo-66) that is common to all isoforms (Prinjha et al. 2000, Chen et al. 2000).
Figure14: Structure of the three Nogo isoform.
All the three isoforms have a common inhibitory domain, Nogo66. Among them, Nogo-A is the only to possess a second inhibitory domain, Amino-Nogo.
The Nogo receptors
In the early 2000’s using a soluble form of Nogo-66, Strittmatter and colleagues cloned a binding partner for Nogo-66 which they called…Nogo receptor (NgR) (Fournier et al. 2001). This latter is a GPI-linked protein and was found to be expressed on the surface on main neuronal types of the CNS (Josephson et al. 2002). Remarkably, NgR has been shown to interact with high affinity to MAG (Domeniconi et al. 2002) and OMgp (Wang et al. 2002b), which are also myelinic inhibitors of the regeneration, but nonetheless structurally different from Nogo suggesting a common pathway for the inhibition of the regeneration.
However NgR lacks an intracellular domain and by consequence cannot transduce the myelin-
However NgR lacks an intracellular domain and by consequence cannot transduce the myelin-