Myelin in the CNS

Myelin is a multilayered glial membrane produced by oligodendrocytes, a structural cell type (glia) present in the central nervous system (CNS). A functioning CNS is dependent upon myelin for fast, saltatory impulse propagation between neurons [1]. The myelin-axon structure, with interspersed unmyelinated axonal segments (nodes of Ranvier) and myelinated axonal segments (internodes), allows the electrical impulses to jump from node to node [1]. This rapid saltatory conduction is achievable due to this alternating myelin-axonal structure whereby the node is excited and rapidly depolarizes at the next node as a consequence of the highly resistant myelin sheath at the internode [2]. Myelin is required for normal brain function. As such, it is well appreciated that defects in myelin or in myelination are associated with psychiatric disorders, such as schizophrenia and depression [1].

Myelination and Plasticity

Although the function of myelin is shared between all vertebrates, the formation of myelin differs between species. Myelination is the process of axon ensheathment by oligodendroglial membranes and occurs in a hierarchical, caudal to rostral progression primarily during postnatal maturation. The temporal progression [2] and extent [3] of myelination in the primate brain is markedly different to that of rodents, with the primate CNS being more heavily myelinated. In humans, myelination commences at birth at the occipital pole, gradually progressing towards the frontal lobe [4, 5]. This occurs most rapidly in the first two postnatal years and slows down thereafter continuing well into adulthood. In contrast, myelination of the rodent CNS commences at 10 days postnatal reaching a peak at postnatal day 20 [2]. Myelination in both rodent and human brains progresses in a caudal to rostral direction [6, 7]. Sensory areas are typically myelinated before motor areas, with long-range projections preceding local connectivity [6, 8].

Developmental plasticity in the early postnatal period results in a high potential for changes to neural networks [9], which does not persist into adulthood once stable myelinated connections are formed. A low level of plasticity is retained in the adult CNS where some oligodendrocyte precursor cells do continue to proliferate and differentiate, allowing myelin to be slowly but continuously remodeled in discrete regions, such as sensory and motor cortical areas, where refining certain connections is important [1]. This level of plasticity allows for functional activity to trigger myelination in order to strengthen connections related to the learned or repeated tasks being performed.

Myelin-Associated Proteins

Myelinating oligodendrocytes express three protein groups that make up attractive/repulsive oligodendrocyte surface molecules: the myelin-associated inhibitors (MAIs), classical guidance molecules (semaphorins, ephrins, and netrins), and chondroitin sulfate proteoglycans (CSPGs). These proteins affect oligodendrocyte migration and differentiation and help to prevent aberrant myelination. For example, they limit the formation of internodes to prevent pathological double myelination of already ensheathed axons or myelination of the nodes of Ranvier [1]. Interestingly, these growth-promoting and/or -restricting neurotrophic and axonal guidance factors are not only critical during CNS development but also play key roles in regulating synaptic plasticity and learning, and in repair inhibition following injury in the adult CNS. The MAIs in particular have become a major focus in the field of neuroplasticity and repair where blocking their activity has the potential to improve functional outcomes following CNS disease and/or injury.

Myelin-Associated Inhibitors (MAIs)

Schwab and Caroni [10] identified mature oligodendrocytes and CNS myelin as nonpermissive substrates for neurite outgrowth [11, 12]. Since this discovery, the components of myelin responsible for this neurite outgrowth inhibition have been identified and collectively termed the MAIs. The MAIs are a group of membrane-bound proteins that have roles in CNS development, plasticity, and regeneration inhibition following injury. MAI ligands include neurite outgrowth inhibitor A (NogoA), myelin-associated glycoprotein (MAG), and oligodendrocyte-myelin glycoprotein (OMgp). These ligands act through two primary receptors: Nogo receptor 1 (NgR1) and paired immunoglobulin-like receptor B (PirB) (Fig. 1a) [13–16]. MAG can alternatively signal through NgR2—a structural homolog of NgR1 (Fig. 1a) [17]. The MAI ligands signal through these various receptors to achieve the same downstream effect by converging on a common downstream pathway allowing for compensatory function and a level of redundancy if signaling through one is inhibited/restricted.

Fig. 1
figure 1

Schematic of myelin-associated inhibitor ligand-receptor interactions. a Neurite outgrowth inhibitor A (NogoA), myelin-associated glycoprotein (MAG), and oligodendrocyte-myelin glycoprotein (OMgp) can collectively signal through Nogo receptor 1 (NgR1) and paired immunoglobulin-like receptor B (PirB). MAG preferentially signals through a structural homolog of NgR1 known as NgR2. These interactions ultimately lead to actin depolymerization and subsequent restriction of neurite outgrowth. b NiG∆20 can signal through sphingosine-1 phosphate receptor 2 (S1PR2). NiG∆2 signals through an unknown receptor. NogoA-24 and Nogo-C39 are able to bind to NgR1 and PirB alongside Nogo-66 to help mediate a high affinity interaction. c Due to the lack of an intracellular domain, signaling through NgR1 is dependent on the recruitment of co-receptors: leucine rich repeat and immunoglobulin-like domain-containing NgR interacting protein 1 (LINGO-1) and p75 or TAJ/TROY. The NgR1/co-receptor complex is stabilized by gangliosides GD1a and GT1b

Full size image

MAI ligand-receptor interactions are repulsive to neurite outgrowth and are generally associated with repair inhibition following CNS injury. However, MAIs also play critical roles during CNS development as repulsive guidance cues affecting neuronal migration, myelination, and axonal target specificity. More recently, the MAI proteins have been shown to play important roles in restricting both developmental and adulthood plasticity in the visual and sensorimotor cortices.

Neurite Outgrowth Inhibitor A (NogoA)

NogoA is a transmembrane protein of 1192 amino acids encoded by the reticulon-4 gene, with a highly conserved 188 amino acid reticulon homology domain (RHD) at the C-terminus (Fig. 1b) [18, 19]. The N-terminal domain (amino-Nogo) distinguishes NogoA from its family isoforms NogoB and NogoC. NogoA comprises two hydrophobic transmembrane domains separated by a conserved extracellular 66 amino acid loop (Nogo-66). Within Nogo-66, the two regions comprised of amino acids 1–33 and 36–41 are instrumental for binding to NgR1 and activating downstream signaling, respectively [19, 20]. Both amino-Nogo and Nogo-66 induce growth cone collapse and inhibit neurite outgrowth, albeit through different receptors [19]. It is important to note that only the Nogo-66 domain of NogoA, and not the other family isoforms, possesses strong inhibitory activity towards neurite outgrowth, signaling through NgR1 [15]. Two regions termed NiG-∆20 and NiG-∆2 (Fig. 1b) have been identified as mediating the inhibitory functions of amino-Nogo, the former signaling through sphingosine-1 phosphate receptor 2 (S1PR2) (Fig. 1a) and the latter through an unknown receptor [21–23]. Notably, regions within both the amino-Nogo and C-terminal segments of NogoA, termed NogoA-24 and Nogo-C39 (Fig. 1b), respectively, have been identified as important non-inhibitory components of NogoA that are able to bind NgR1 and PirB with high affinity (Table 1) [21, 24]. Interestingly, NogoA can exhibit several different topologies, where evidence suggests that the typical intracellular amino-Nogo and extracellular Nogo-66 is reversible with amino-Nogo also detectable on the extracellular surface and Nogo-66 also able to face intracellularly [18, 23, 25–27]. Since NogoA can exhibit a topology where amino-Nogo, Nogo-66, and the C-terminus of NogoA are extracellular, NogoA-24 and Nogo-C39 may simultaneously bind to the receptor alongside Nogo-66 to create a trivalent ligand leading to downstream neurite outgrowth inhibition through NgR1 or PirB. This tri-valency may explain why NogoA is the most potent neurite outgrowth inhibitor of the three MAI ligands being able to outcompete MAG and OMgp in binding their receptors (Table 1) [28]. Consequently, these aforementioned regions and varying topology make NogoA capable of high affinity interactions with various receptor combinations in order to induce potent neurite outgrowth inhibition.

Table 1 MAI ligand-receptor dissociation constants. Approximate dissociation constant (Kd) values highlighting binding affinity between MAI ligands and their receptors expressed in nM [20, 24]
Full size table

In the developing and adult rodent brain, neuronal NogoA is associated with areas of high plasticity, such as the olfactory bulb, hippocampus, and spinal cord [29–33]. NogoA is also expressed at the level of neuronal synapses [34–36]. It is important to note that the role of neuronal NogoA in the adult CNS is largely unknown. In the mature CNS, NogoA is present on the surface of myelinated axons as well as oligodendrocyte somas and processes [36]. As a reticulon protein, NogoA is additionally expressed in the endoplasmic reticulum (ER) [19], where it is thought to help maintain the tubular structure of the ER adopting a topology where Nogo-66 is typically facing the ER lumen and amino-Nogo is facing the cytosol [37]. In this form, NogoA has been shown to be required for ER tubule formation in vitro, highlighting an essential role in intracellular trafficking [37].

Less is known about the shorter Nogo isoforms. NogoB is ubiquitously expressed in a wide range of tissues and is thought to be involved in vascular remodeling and apoptosis [38]. NogoC is mainly expressed in skeletal muscle and to a lesser extent in the CNS, but so far its function remains unknown [25]. For a more comprehensive review on the functions of Nogo proteins see [39].

Myelin-Associated Glycoprotein (MAG)

MAG is a transmembrane protein of 626 amino acids containing five immunoglobulin (Ig)-like domains in the extracellular region and was the first identified MAI expressed by myelinating glia (Fig. 1a) [40]. MAG comprises two isoforms: large (L)-MAG and small (S)-MAG. S-MAG is the predominant form found in the peripheral nervous system (PNS) and L-MAG in the CNS [40, 41], with the latter abundantly expressed at the initiation of CNS myelination [42]. The ligand is localized to periaxonal oligodendroglial membranes but is a minor component of myelin [43]. Interestingly, MAG appears to have a dual role in axonal growth, which is dependent upon the developmental age of the neurons with which it interacts. MAG signaling promotes neurite outgrowth in immature/young neurons [44, 45] but conversely inhibits mature/older neurons [46–48]. This is a very interesting trait as it is the only MAI ligand to promote neurite outgrowth during CNS development. MAG also has an important association with gangliosides, which are sialic-acid-bearing glycosphingolipids involved in axon-myelin stability [49]. For a comprehensive review on the roles of gangliosides see [50].

Oligodendrocyte-Myelin Glycoprotein (OMgp)

OMgp is a glycophosphatidylinositol (GPI)-anchored protein comprising 440 amino acids with a series of tandem leucine-rich repeats (LRRs) and a short cysteine-rich motif (Fig. 1a) [51–53]. It is a minor glycoprotein found in CNS myelin and is not expressed in any other tissue outside the brain and spinal cord [51]. OMgp is found on the cell membranes of myelinating oligodendrocytes and the glial-axonal interface of myelinated axons [54–56]. The spatiotemporal expression of OMgp coincides with the caudal to rostral progression of CNS myelination [53]. In addition to its expression in white matter, OMgp is abundantly expressed by adult CNS neurons, especially large projection neurons [54] where its role is unknown.

Myelin-derived OMgp was initially found to inhibit neurite outgrowth in cerebellar, hippocampus, dorsal root ganglion, and retinal ganglion neurons [55]. Interestingly, OMgp-null mice show collateral sprouting at nodes of Ranvier, indicating that OMgp has a critical role in maintaining the integrity of the myelin-axon interface [57].

MAI Receptors

Nogo Receptor (NgR)

NgRs are encoded by the reticulon-4R gene and include NgR1, which is the most widely studied of the NgRs, and two structural homologs known as NgR2 and NgR3. NgR1 was first identified in 2001 by Fournier et al. [15] and is a protein of 473 amino acids made up of a signal sequence followed by eight LRR domains, a cysteine-rich LRR-like carboxy terminal (CT) flanking domain, a unique C terminal and a GPI anchor. NgR1 is not expressed on all mature neurons but has been reported in discrete regions of the neocortex, hippocampus, pons, and cerebellum [15, 58]. Additionally, prior data suggests that although NgR1 is required for acute growth cone collapse, other MAI receptors, such as PirB, may play more of a role in chronic growth inhibition [59].

The LRR domains are essential for NgR1 interaction with Nogo-66, and to the MAG and OMgp receptor-binding sites [60]. Recent studies have demonstrated that the CT flanking domain is additionally required for inducing the neurite outgrowth inhibitory effect of these ligand/receptor complexes [15]. Due to a lack of an intracellular domain, signal transduction through NgR1-MAI interactions is dependent on the formation of signaling complexes with several transmembrane co-receptors (Figs. 1a, c and 2; NgR1 pathway) [61–64]. Like NgR1, NgR2, the receptor through which MAG preferably signals, lacks a signaling domain and interacts with unknown co-receptors (Fig. 2; NgR2 pathway) [17].

Fig. 2
figure 2

Schematic of neurite outgrowth inhibitory signaling pathways. The neurite outgrowth (Nogo)-receptor 1 (NgR1) signaling pathway includes RhoA, Rho kinase (ROCK), LIM kinase (LIMK), and Cofilin. The paired immunoglobulin-like receptor B (PirB) signaling pathway occurs through three interconnected downstream effectors. Activation of PirB can subsequently activate plenty of sarcoma (src) homology domain 3 (SH3) POSH though an unknown mechanism. Activation of POSH either diverts the signaling pathway back to ROCK by activating Shroom3, or activates a kinase cascade via leucine zipper kinase (LZK), mitogen-activated protein (MAP) kinase kinase (MKK), Jun N-terminal kinase (JNK) and Myosin II. Alternatively, POSH activation via PirB phosphorylation can recruit src homology protein 1/2 (SHP1/2), which dephosphorylates tyrosine kinase B (TrkB), leading to a MAP kinase (MAPK), Rac/Rho, and Cofilin signaling cascade. All these pathways ultimately lead to actin disassembly, thus restricting neurite outgrowth. The NgR2 and sphingosine-1 phosphate receptor 2 (S1PR2) signaling pathways remain unclear but similarly result in neurite outgrowth inhibition and destabilization of synapses, respectively

Full size image

Two types of co-receptor complexes have been demonstrated as essential in inducing the downstream signaling pathways resulting from NgR1-MAI interactions. LRR and Ig-like domain-containing NgR interacting protein 1 (LINGO-1), a well-recognized negative regulator of oligodendrocyte precursor cell (OPC) differentiation [65], must either form a complex with NgR1 and the neurotrophin receptor (NTR) p75, or with TAJ/TROY, an orphan tumor necrosis factor (TNF) receptor family member (Fig. 2 NgR1 pathway) [61–64, 66]. Interestingly, the interaction between MAG and certain gangliosides, outlined above, can affect the formation of co-receptor complexes. Interaction of MAG with the gangliosides GD1a and GT1b is involved in the recruitment of p75 to the receptor-signaling complex [67]. Additionally, it has been demonstrated that the ganglioside GT1b is important for the interaction between NgR1 and Lingo-1 [68]. It is evident that these gangliosides are critical to the formation and stability of the NgR1/Lingo-1/p75-signaling complex and for subsequent signal transduction. The p75 (NTR)-inclusive complex is thought to be important for MAI signaling during CNS development as it is highly expressed during periods of axon outgrowth and dendritic arborization, with decreased expression over the postnatal period [61]. In adulthood, only a few subpopulations of mature neurons express p75 [61, 69–71]. Furthermore, although adult mice lacking p75 show decreased neurite outgrowth inhibition, a significant amount of inhibitory potential remains, and these mice fail to show any significant regenerative potential following CNS injury [63, 72]. On the other hand, TAJ/TROY is widely expressed in the adult CNS [73] and interference of TAJ/TROY function is able to prevent MAI-induced neurite outgrowth inhibition [61] presumably by substituting for p75 as a co-receptor for NgR1/Lingo-1 in the adult CNS. For a more comprehensive review on NgR signaling complexes see [74].

Paired Immunoglobulin-Like Receptor B (PirB)

PirB, a mouse ortholog to human leukocyte Ig-like receptor B2 (LILRB2), is a major receptor for NogoA, MAG, and OMgp [13]. As an MHC class 1 receptor, PirB was first described as an agonist to integrin receptors in the immune system containing a single transmembrane domain with three-immunoreceptor tyrosine-based inhibitory motifs [75]. PirB is mainly expressed by astrocytes and subsets of neurons in the CNS and has been detected in the cerebral cortex, hippocampus, cerebellum, and olfactory bulb [13, 76–78].

Rodent studies have demonstrated high affinity interactions between PirB and all three MAIs, which induces growth cone collapse (Fig. 1a) [13]. MAI signal transduction through PirB is dependent upon the interaction of its intracellular domain with the protein tyrosine phosphatase, sarcoma (src) homology protein 1/2 (SHP1/2) (Fig. 2 PirB pathway) [78]. Recent evidence has additionally suggested that p75 interacts with PirB upon MAI ligand binding and is essential to the activation of SHP1/2 [79]. These studies implicate p75 as an important component of both the NgR1 and PirB signaling complexes through which all the MAIs are able to induce neurite outgrowth inhibition.

MAI Roles in CNS Development

Although first identified as an inhibitor of neurite outgrowth in the mature CNS, NogoA also has several developmental roles. The ligand is abundantly expressed by both PNS and CNS neurons [29, 80] and on radial glia [81, 82] prior to myelination. Genetic ablation of NogoA is deleterious to both radial [83] and tangential [81] neuronal migration, resulting in delayed corticogenesis and demonstrating it has a critical role in neuronal migration. NogoA is also implicated in guiding uncrossed axons to the ipsilateral optic tract during the development of the mouse optic chiasm [84]. The genetic deletion or neutralization of NogoA in mouse-derived dorsal root ganglion (DRG) explants results in increased neurite outgrowth and fasciculation coupled with decreased branching of cultured DRG neurons [85]. These data suggest that NogoA expression on neurons and radial glia is a repulsive cue that can act on growing fibers contributing to their migration, guidance, and regulation during CNS development.

MAG signaling is important for the initiation and maintenance of myelination and contributes to oligodendrocyte maturation during CNS development [28, 42, 86]. MAG-null mice exhibit abnormalities that include myelin outfolds, multiple myelin rings, uncompacted myelin sheaths, and lack or excess of cytoplasm in the periaxonal space [86]. These structural abnormalities result in overproduction of myelin, delays in myelination, and accelerated axonal degeneration over time. These data suggest that MAG is an important structural component of myelin and is crucial for appropriate myelination of neural circuitries.

OMgp’s precise physiological role is relatively unclear [54, 87]. Expression of OMgp has been reported prior to the onset of myelination, with peak expression in late stages, and has been implicated in oligodendrocyte proliferation and differentiation [28, 87, 88]. Initial evidence suggested that OMgp aggregated at the nodes of Ranvier, further indicating an important role in myelination [57, 89]. However, a more recent study has invalidated these previous findings concluding that OMgp is not found in, nor is essential for paranodal architecture [90]. It is evident that more work remains to be done in order to understand the normal function of OMgp in CNS myelination. A recent study has also suggested that OMgp may contribute to axon target specificity. This was demonstrated by the over extension of thalamocortical axons in the barrel field of OMgp-null mice past layer 5 into the more superficial layers [87]. These data suggest that OMgp plays important roles in axonal pathfinding during CNS development, whereas its role during CNS myelination is unclear.

For a summary of NogoA, MAG and OMgp roles during CNS development see Table 2.

Table 2 Role of MAI ligands and receptors during CNS development
Full size table

MAI Signaling in CNS Plasticity

In addition to the involvement of the MAIs during CNS development, MAI signaling has more recently been implicated in the regulation of CNS plasticity. NogoA, OMgp, NgR1, and PirB have been detected at pre- and post-synaptic terminals [36, 78, 87], indicating that they may interact with one another to negatively regulate synaptic plasticity [94]. Moreover, high levels of NgR1 expression have been detected in areas of increased plasticity, such as the hippocampus [58]. NgR1 has functionally been demonstrated to be involved in the restriction of ocular dominance plasticity in the developing mouse visual cortex [91, 92], where deletion resulted in the extension of the “critical period” of up to four times as long as wild-type mice.

PirB has been shown to actively limit the extent of experience-dependent plasticity throughout life. Blockade of PirB promoted the formation of new synapses, restoring visual acuity and spine density following monocular deprivation [93]. Lack of PirB resulted in increased ocular dominance plasticity both during and outside of the critical period [78]. Monocular deprivation studies following blockade of NgR1 or PirB have demonstrated improved structural plasticity, with synaptic connectivity altered mostly at the postsynaptic level [93, 95].

NgR1- and PirB-dependent interactions with NogoA are effective in restricting plasticity in the visual and sensorimotor cortices and hippocampus [49]. Following monocular deprivation, NogoA knockout (KO) mice exhibited higher desegregation of retinogeniculate projections compared to wild-type mice, where a greater number of ectopic terminals from the contralateral eye innervated into the ipsilateral territory [96]. In addition, increased long-term potentiation, spine density and improved learning of skilled forelimb-reaching tasks were evident after functional neutralization of NgR1/NogoA in rats [97]. These studies provide evidence that NgR1- and PirB-dependent NogoA signaling is important in restricting both the plasticity of the retinogeniculate pathway and the sensorimotor cortex.

For a summary of NgR1 and PirB roles during CNS development see Table 2.

Neurite Outgrowth Inhibition Pathways

One of the common aspects of the various growth inhibitory factors, including the MAIs, is the convergence on similar downstream pathways involved in actin cytoskeletal dynamics. The most common of these is the NgR1/RhoA/Rho Kinase (ROCK) signaling pathway (Fig. 2; NgR1 pathway), which is responsible for regulating actin cytoskeletal dynamics in neurites [98]. RhoA is a member of the Rho guanosine triphosphate (GTP)-ases, which is primarily responsible for signal transduction of axonal guidance molecules during development. RhoA is activated by the displacement of a guanine dissociation inhibitor [99, 100]. This allows a guanosine exchange factor to facilitate the phosphorylation of guanosine diphosphate to GTP, thereby activating ROCK, which subsequently phosphorylates LIM kinase (LIMK), which in turn phosphorylates Cofilin, resulting in actin depolymerization (Fig. 2; NgR1 pathway) [101]. NgR2 interacts with unknown co-receptors and its downstream signaling pathway is unclear (Fig. 2; NgR2 pathway). NgR1/2 signal transduction both result in neurite outgrowth inhibition and retraction.

PirB signals through three downstream effectors that can crosstalk (Fig. 2; PirB pathway). The first involves the activation of plenty of src homology domain 3 (POSH), via PirB, by an unknown mechanism [102]. POSH subsequently activates Shroom3, which either converges with the ROCK pathway (Fig. 2; NgR1 pathway), or activates leucine zipper kinase (LZK), initiating a secondary downstream kinase cascade, resulting in restriction of axon outgrowth (Fig. 2; PirB pathway) [103]. Phosphorylation of PirB also recruits SHP1/2, which activates the third effector, a tyrosine kinase (Trk)/mitogen-activated protein kinase (MAPK) pathway, by dephosphorylating TrkB [104]. The result is a signaling cascade ending in neurite outgrowth inhibition via actin disassembly (Fig. 2; Trk pathway) [103]. The NgR1 and PirB signaling pathways exhibit several common signaling and downstream components. This means that there is high potential for cross talk and even compensation if one pathway is impeded in some way. The importance of these two actin cytoskeletal-related signaling pathways cannot be understated as they ultimately lead to neurite outgrowth inhibition and destabilization of synapses affecting both normal and pathological CNS function.

The sphingosine-1 phosphate receptor 2 (S1PR2) pathway, through which NiG∆20 signals, is also not well understood, but ultimately leads to the destabilization of synapses (Fig. 2; S1PR2 pathway) [22].

MAI Links to Neurodegenerative Diseases

In addition to their roles in CNS development and plasticity, the MAIs have also been associated with neurodegenerative diseases. It has been demonstrated that NogoA is able to bind and inhibit the two anti-apoptotic proteins Bcl-XL and Bcl-2 [105], implicating a pro-apoptotic role. This may contribute to neurodegenerative pathology as a consequence of dysregulation of the microenvironment. NogoA has also been linked to neurodegenerative diseases, such as multiple sclerosis (MS) [106] and amyotrophic lateral sclerosis [107], its presence affecting their progression and severity.

Studies have also linked MAG signaling to neuroinflammation and demyelinating neuropathies, such as MS, where patients develop significant levels of MAG autoantibodies [108]. Several of these studies have shown that MAG signaling may mediate axon stability and integrity, and afford a level of neuroprotection to axons under pathological conditions [109].

More recently, NogoA and NgR1 have been implicated in the amyloid precursor protein (APP) pathway. Under the relevant conditions, APP cleavage leads to the formation of amyloid beta (Aβ) fragments, which precede the degenerative pathology characteristic of Alzheimer’s disease (AD). Moreover, APP processing can occur in lipid rafts [110], where the MAI ligands and receptors typically signal with one another. NogoA is able to bind to the β-secretase (BACE1) and NgR1 is able to bind to APP [111], in combination likely acting as competitive inhibitors to APP processing. Studies have shown that overexpressing NogoA/NgR1 decreases Aβ deposition and that deleting NgR1 results in elevated levels of Aβ [112, 113]. It appears that NogoA may sequester BACE1 processing of APP and that NgR1 may block APP processing by β- and α-secretases.

Consequences of CNS Injury

Injury to the CNS can occur as a result of exogenous factors, such as physical force as is the case with traumatic brain injury (TBI) and spinal cord injury (SCI), or as a result of endogenous factors, such as a cerebrovascular clot, as is the case with hemorrhagic and/or ischemic stroke. In the minutes following the initial CNS injury (acute phase), the irreversible death of a large number of neurons occurs rapidly at the site of injury (lesion core) as the direct result of the trauma [114–116] or from excitotoxicity due to excessive release of glutamate [117]. Cell swelling, mitochondria, and endoplasmic reticulum breakdown occurs leading to rupturing of plasma and intracellular membranes [118]. Disruption of the blood brain barrier as a direct result of the injury and the accumulation of free radicals in the lesion core causes local and acute tissue injury and leads to the chemotactic recruitment of immune (peripheral leukocytes and microglia) and astrocytic cells to mitigate the extent of damage by removing harmful debris [115, 119]. During this inflammatory response, microglia and astrocytes secrete both proinflammatory cytokines and neuroprotective factors [120–122].

Within hours of the initial CNS injury (sub-acute phase), cells are in a state of functional inactivity with metabolic processes continuing [117]. This zone is known as the penumbra or peri-lesion area that surrounds the lesion core. In the case of stroke, the penumbra may experience drastically reduced blood flow [123], and cells may undergo a second wave of apoptosis. In ischemic stroke, the secondary injury can occur as a result of persisting ischemia and hypoxia, or following reperfusion from oxidative (free radicals), excitotoxic (excessive glutamate), and osmotic (cerebral edema) stress [124–126]. These stresses, coupled with axonal degeneration, severely affect glial populations, especially in terms of astrocytic recruitment and reactivity, and oligodendroglial integrity [127–129]. Damage to neuronal processes, including loss of terminal fields and the subsequent anterograde degeneration that follows can also result in neuronal apoptosis in areas distal to the lesion core, expanding the size of the penumbra [130]. Notably, unlike the lesion core, up to half of the penumbral region may be salvageable, i.e., the damage is thought to be reversible [114, 130, 131]. However, the secondary wave of damage does compromise both the lesion core and penumbra leading to protracted recruitment and accumulation of microglia and astrocytes, as well as the continued death of neuronal and oligodendrocyte populations [129], thereby exacerbating the hostile environment within these areas.

In the uninjured CNS, oligodendrocytes and their glial membranes (myelin) form a physical protective barrier for axons and facilitate neuronal function. However, the loss of terminal fields and subsequent retraction of axons following injury results in a breakdown of the myelin sheaths, which accumulate as myelin debris exposing axons to the extracellular environment. This directly affects the oligodendroglial-neuron microenvironment, which impacts on the expression of oligodendrocyte surface molecules, such as the MAIs, ultimately leading to cytoskeletal changes within the axon. In addition, the recruitment of reactive astrocytes and subsequent CSPG deposition leads to the formation of a glial scar, which further suppresses neural regeneration [132, 133]. It is important to note that NgR1 and NgR3 are also receptors for CSPGs, which comprises the major component of the extracellular matrix in the glial scar following CNS injury [134]. As such, the resulting environment becomes extremely refractory to neurite outgrowth and prevents re-innervation of the damaged area.

MAI Response to CNS Injury

The effects of MAI signaling on neural repair inhibition have been extensively studied following adult injury in rodent models of TBI, SCI, and stroke. However, the effects of CNS injury on the MAIs and their receptors themselves, i.e., the molecular changes that occur as a consequence of CNS injury have not been well documented.

Following CNS injury, the MAIs and their receptors are thought to play key roles in repair inhibition. The suggestion is that the damaged axon-myelin interface causes molecular changes within the cells resulting in upregulated expression of the MAIs on neurons and oligodendrocytes. This elevated expression results in axon-myelin and axon-axon repulsion mediated through the MAI receptors affecting re-myelination and re-innervation at the injury site. Furthermore, the breakdown of the myelin sheath and subsequent accumulation of myelin debris results in exposure of the MAIs to the extracellular environment. This free myelin is presumably able to interact with surrounding neurites expressing the MAI receptors, thereby causing repulsion. Ultimately, this repulsive environment that surrounds the injury site has high potential to repel axons that attempt to re-form functional connections to the damaged area.

However, the current evidence from the published literature surrounding the changes in MAI ligand and receptor expression in various CNS injury models is inconsistent and is likely due to inter-/intra-species differences and the specific injury paradigms employed.

TBI and SCI

The published inconsistencies surrounding MAI ligand-receptor expression following CNS injuries are most evident in the field of TBI and SCI research. A report investigating cortical lesions (longitudinal cuts) in rats observed no change in the expression of NogoA in the gray or the white matter from 1 day up until 28 days post injury (dpi) aside from a slight upregulation of NogoA at 4dpi in cortical neurons [29]. However, fluid percussion injury (a form of TBI) in the rat parietal cortex revealed an increase of NogoA on neurons and oligodendrocytes in both gray and white matter structures between 1 and 7 dpi. Surprisingly, NgR1 expression was not altered [135]. The differences in MAI ligand and receptor expression here seem to be related to the specific model of TBI. In the case of the cortical lesions involving longitudinal cuts, only local MAI expression was affected, but the percussive cortical injury affected both local and distal MAI expression. In the latter model, the injury likely penetrated into the white matter resulting in those cell populations additionally responding to the injury.

In a rat SCI model, no change in NogoA was detected around contusion sites [136]. Similarly, NogoA upregulation was not observed at the transection sites of induced rat spinal cord lesions [29]. Other studies have reported increases in OMgp and NgR1 following rat SCI [137], and decreases in NogoA in central parts of spinal cord lesions, but increases at the borders of these lesions [30, 36]. As with the aforementioned TBI models, the variable reports in MAI ligand and receptor expression following SCI are likely a consequence of different injury paradigms and could be overcome by normalizing the injury models.

Stroke

In a global ischemic stroke model [138], a transient increase in NogoA expression was observed localized to the myelin sheath, reaching peak levels between 24 and 48 h in rats, and returning to baseline by 96 h. Interestingly, the researchers also observed long-lasting elevation of NgR1 and RhoA up to 7 days post-stroke, which differs from the findings of the aforementioned TBI lesion studies. A similar study revealed no change in NogoA expression in white matter structures, but an upregulation of NogoA in cortical neurons distal to the lesion site with a peak expression at 28 days following a middle cerebral artery occlusion (MCAO) model of stroke in rats [139]. The group also observed a downregulation of NogoA in the peri-lesion cortex, and no significant changes in white matter structures following stroke. These studies have overlapping time points and yet observed different NogoA responses to the induced stroke. As is the case with the TBI and SCI studies, normalizing stroke injury models would likely reveal more consistent results.

In one of the few MAI-related nonhuman primate (NHP) studies performed, increased levels of NogoA were found on oligodendrocytes in white matter structures adjacent to the lesion core up to 2 months following permanent MCAO in adult marmoset monkeys (Callithrix jacchus) [140]. However, this study discounted cells with neuronal morphology as few were seen in the gray matter. This study indicates that there are likely differences between MAI responses between species, which is important when thinking about translational research.

Finally, the receptor PirB has also been observed to be markedly elevated following MCAO in both mice and rats at various time points, where mRNA and protein levels remain elevated between 2 h and 7 days after reperfusion, with expression peaking at 24 h [76, 141]. However, one issue with these results is that mouse PirB only has 50% homology to the human ortholog LILRB2 [142]. Such a large difference in protein structure between species could affect the function of the receptor. Furthermore, LILRB2 transcripts can differ in length due to numerous isoforms, which could potentially lead to functional redundancy when removing or blocking the protein in humans compared to mice.

The expression of MAIs and their receptors following CNS injury is inconsistent within and between injury models and species. Additionally, where there are large differences in protein structure between species, as is the case with PirB in rodents versus primates, the results become questionable. This ambiguity in the literature begs the question of whether these CNS injury models are translatable to humans, and whether we can use them to gauge what is happening at the molecular level. In terms of clarity, studies investigating the interference of MAI signaling pathways are more convincing.

Towards Translatable Therapies

The prevailing hypothesis is that interfering with MAI signaling pathways may improve regenerative capacities following CNS injury. Several studies have successfully demonstrated increased neurite outgrowth resulting in improved functional outcomes in rodent injury models after blocking MAI ligand-receptor interactions.

Neutralizing Antibodies

Antibodies that block MAG [143] and NogoA [144–149] from binding NgR1 have previously demonstrated efficacy in rodent stroke recovery. In one of the latter experiments [147], improved neurological deficits and increased neuronal plasticity were observed when administering anti-NogoA immunotherapy 9 weeks following stroke. This demonstrates that blocking NogoA can afford benefits in the chronic stages following cerebral ischemia. Furthermore, neutralizing NogoA in the very early stages (hyper-acute phase) following stroke, when the potential for persisting or recurring ischemia is high, may result in increased mortality and exacerbated neurological deficits [150]. This indicates that the timing for any therapeutic targeting NogoA is critical. Alternative routes of administration have been explored for anti-NogoA therapies, including DNA vaccination techniques, which generate a tolerable antibody response even during severe inflammation [151]. Improved functional outcomes have also been observed when blocking more specific components of the MAI signaling pathway, such as Lingo-1. Blockade of Lingo-1 in adult rats following acute SCI via continuous intrathecal infusion of Lingo-1 polyclonal antiserum over 4 weeks resulted in decreased RhoA activation, a reduction in neuronal apoptosis, and improved hindlimb function [152]. A subsequent study demonstrated similar results using a truncated version of Lingo-1 [153].

Competitive Inhibition

Exogenous inhibition of NgR1 with peptide antagonists has previously been shown to mitigate axonal damage, enhance axonal sprouting and improve motor function following cortical injury in rodents [154, 155]. A truncated peptide derived from the Nogo-66 loop called nogo extracellular peptide 1–40 (NEP1–40) was able to promote axon regeneration and improve functional outcomes following SCI, most likely through competitive inhibition of ligand-receptor interaction [156–158]. Exogenous infusion of NEP1–40 in combination with inosine, a purine nucleoside that activates neuronal intrinsic growth potential, was hypothesized to be an enhanced treatment option to promote regeneration following CNS injury [159]. Rats that underwent unilateral stroke in the forelimb motor cortex area and received NEP1–40 + inosine treatment showed increased neurite outgrowth. These combination-treated animals achieved double the number of interhemispheric axon branches projecting into the denervated side of the spinal cord compared to either treatment alone. In addition, this treatment afforded significant recovery of forepaw skilled reaching back to preoperative levels [159].

Genetic Deletion

Cultured myelin taken from MAG- and OMgp-null mice showed levels of axonal outgrowth inhibition comparable to wild-type myelin, but NogoA-null mice showed significantly reduced inhibition of axonal outgrowth in vitro [28]. A triple KO of the three MAI ligands revealed an even greater reduction in inhibition of axonal outgrowth in cultured white matter. These data suggest a synergistic role for MAG and OMgp. Furthermore, when triple KO mice were subjected to SCI, rostral corticospinal tract sprouting and regeneration across the lesion site was observed, which resulted in improved locomotion compared to wild-type mice [28]. However, not all injury models have beneficial outcomes following genetic deletion of MAIs. NogoA/B KO resulted in increased degeneration of motor axons in a mouse model of amyotrophic lateral scleosis [160]. Genetic deletion of NogoA before MCAO induction in adult mice resulted in more severe neurological deficits than in wild-type mice [150]. Furthermore, suppression of NogoA signaling in retinal ganglion cells following optic nerve crush resulted in decreased neuronal survival [161]. These studies indicate that NogoA may have roles in neuronal survival in the injured CNS. Interestingly, both NgR1 and PirB-null mice do not show increased functional recovery following CNS lesions [77, 162, 163]. This may be a result of functional redundancy due to crosstalk between their downstream pathways (Fig. 2 NgR1/PirB pathways).

While there are a few studies that describe potential neuroprotective roles for NogoA [150, 161] and MAG [108, 109] under pathological conditions, the overwhelming data suggests that blocking MAI ligand-receptor interactions can result in increased neurite outgrowth and afford a level of functional recovery for sensorimotor deficits following CNS injuries. This makes the MAI signaling pathways attractive targets for therapeutic intervention.

MAI Blockade: from Bench to Bedside

Despite an immense research focus on regenerative medicine, injuries of the CNS remain untreatable. Indeed, of the thousands of developed therapeutics trialed in rodent models of CNS injury, especially stroke, there has been no successful translation into the clinic [164]. Over the years, many rodent studies focusing on interfering with NogoA signaling [157, 158, 165–168] have demonstrated the increased potential for recovery following SCI through various strategies.

In one of the few studies that utilized a NHP injury model, Freund et al. [169] administered an anti-NogoA antibody to adult macaque monkeys after unilateral transection of the cervical spinal cord. It was observed that treated monkeys regained motor function, i.e., restoration of manual dexterity. Additionally, subsequent analysis revealed increased corticospinal axonal sprouting both rostrally and caudally to the lesion site. Remarkably, initiation of the treatment in the sub-acute period (1 week) after injury revealed an equal level of functional recovery compared to acutely treated monkeys.

The results from these studies led to phase I clinical trials with NogoA for the treatment of acute paraplegic and tetraplegic patients [170]. Although the study was completed in 2011, the results have yet to be released. There was a report of severe neutropenia following treatment with the antibody (ATI-355) [171], which may have led to discontinuation of testing. In contrast, a Rho-GTPase antagonist called BA-210 (Cethrin) reached a phase I/IIa clinical trial reporting safety and tolerability in acute SCI patients as well as enhanced motor function [172]. Additionally, a recent phase I clinical trial using an anti-MAG antibody demonstrated a trend towards improvement of gait velocity in the treatment groups after the first few days following stroke [173]. The only reported drawback was development of transient anti-drug antibodies in two out of the 25 subjects. These clinical results, in addition to those from animal models, highlight the potential of therapeutic strategies targeting MAI signaling in order to improve functional recovery after SCI and stroke in humans.

Over the past decade, Strittmatter and his group have developed an NgR1 (310)-Fc decoy protein [174]. A rat isoform of this NgR1 decoy protein has shown efficacy in the recovery of rats following SCI, when administered 3 days post injury, and following MCAO, when administered 1 week post injury, both for a period of 28 days [154, 175, 176]. More recently, continuous infusion of a humanized isoform of the NgR1 decoy protein, administered 3 days post injury for a period of 4 weeks, has shown efficacy in the recovery of rats following SCI [177]. This humanized decoy protein holds great promise for translation into the clinic. It is also worth mentioning that this NgR1 decoy protein is also able to reduce Aβ plaques in AD mice [112, 113]. Therefore, this humanized decoy protein has exciting potential as a therapeutic following brain injury and/or for neurodegenerative diseases.

The development and testing of future MAI-based therapeutic strategies should take into account the highlighted inconsistencies, use of appropriate model systems, and lessons learned from failed experimental and clinical trials. Characterizing the MAIs collectively in clinically relevant models to gain an in-depth understanding of their various roles, including those other than neurite outgrowth inhibition, is essential in order to maximize the potential for clinical translatability. One of the first steps should be to identify an MAI-specific therapeutic window following a reproducible, focal injury, to a well-characterized system, where the potential for positive functional outcomes is greatest. Finally, drugs targeting multiple parts of the MAI pathway should be developed, tested, and refined as interfering with only one subdivision is unlikely to produce the desired outcome due to the complex nature of the signaling pathway.