Exploiting mTOR Signaling: A Novel Translatable Treatment Strategy for Traumatic Optic Neuropathy?

Abstract

Retinal ganglion cell (RGC) death and a failure of axon regeneration contribute to the profound visual loss experienced by patients after traumatic optic neuropathy (TON), for which there are no effective treatments. Experimental manipulations of cellular signaling pathways in animal models have demonstrated that neuronal survival and axon regeneration in the mature central nervous system (CNS) are possible, and increased understanding of the molecular basis of prosurvival and regenerative signals has led to the identification of candidate targets for novel therapeutic strategies. The axogenic pathway is activated sequentially, after growth factor/receptor binding, through phosphoinositide-3-kinase (PI3K) and the downstream serine/threonine kinase Akt. Akt is a nodal point for the regulation of growth cone dynamics by glycogen synthase kinase (GSK3β) and axon protein synthesis/RGC survival by the mammalian target of rapamycin (mTOR). The mTOR signaling pathway has a pivotal role in numerous cellular processes. It is active during development, but downregulated in the mature CNS and further suppressed by axonal injury, and experimental upregulation of mTOR signaling promotes RGC survival and axon regeneration after optic nerve crush injury. However, several translational challenges remain, including understanding the regulatory mechanisms of axotomy-induced mTOR and GSK3β signaling, and the disparity between the RGC survival and axon regenerative effects. In this review, we explore the molecular basis of RGC regenerative failure and assess the potential for manipulations of mTOR signaling as a novel translatable treatment for TON.

Traumatic optic neuropathy (TON) is associated commonly with blunt ocular trauma and craniofacial injuries, occurring as either a direct injury to the optic nerve (ON) or indirectly after head trauma sequelae, such as edema, hemorrhage, and concussion. ¹ Mechanical disruption to retinal ganglion cell (RGC) axons results in RGC death, with consequent visual loss. ² Traumatic optic neuropathy occurs in up to 5% of patients with closed head trauma ³ and accounts for nearly 20% of ocular injuries sustained by military personnel in combat. ⁴ Although spontaneous visual improvement may occur in a minority of patients, ⁵ permanent visual loss is the most common outcome. Current clinical evidence provides no indication of the superiority of either steroid treatment or surgical optic canal decompression over conservative management. ^6–8 Indeed, there is evidence that steroids have an unacceptable risk of adverse side effects and complications. ⁹ Amelioration of the long-term morbidity associated with TON with an effective treatment represents an unmet clinical need, and the development of viable therapeutic options for TON may provide ophthalmologists with an opportunity to prevent blindness or even restore sight. 

Unlike the peripheral nervous system (PNS), in which axon regeneration after injury is associated with limited functional recovery, ^10,11 most axons of the adult mammalian central nervous system (CNS) compromised by injury or disease do not regenerate. Optic nerve regenerative failure has been attributed to axotomy-induced apoptotic RGC death, the inhibitory microenvironment of the ON, suboptimal stimulation of axon regenerative responses, and a lack of intrinsic axogenic capacity of mature RGC. ^12,13 However, promoting RGC survival and axon regeneration is not an impossible aim, as specific experimental manipulations have demonstrated that RGC survive and regenerate axons in vivo after injury, for example through peripheral nerve (PN) grafts sutured to the transected end of the ON, ^14–16 and after intravitreal implantation of a PN fragment, ^17,18 lens injury, ^19,20 induction of intraocular inflammation, ²¹ intravitreal delivery of neurotrophic factors, ²² and targeted gene deletion ^23–25 (see Table). The molecular basis of these RGC survival and proregenerative paradigms is becoming increasingly clear, with selected modulation of the intracellular signaling pathways that mediate changes in RGC gene expression emerging as a class of novel therapeutics. 

Table

 

Comparison of Experimental Approaches to Promote RGC Survival and Axon Regeneration

Intervention, Reference	Species	RGC Survival	Axon Regeneration Beyond Lesion	Comment
Lens injury19,20	Rat	∼24% at 2 wk	∼1500–1900 axons at 0.5 mm*
Intravitreal zymosan21	Rat	Increased × 2 vs. control	∼1200 axons at 0.5 mm†	Axon growth up to 4.7 mm
Intravitreal Pam3Cys80	Rat	No increase in survival vs. control in vitro	∼700 axons at 0.5 mm†	2 injections, 7 days apart for max effect
Intravitreal neurotrophic factors22	Rat	Increased × 2 vs. controls in vitro	∼1000 axons at 2 mm*	FGF2/NT3/BDNF transfected fibroblasts
RhoA blockade65	Rat	No increase compared to lens injury alone	∼250 axons at 0.5mm†
RhoA blockade + lens injury65	Rat	No increase compared to lens injury alone	∼1500 axons at 0.5 mm†
PTEN deletion23	Mouse	∼45% at 2 wk	∼1500–1700 axons at 0.5 mm†	Few fibers > 3mm to chiasm by 4 wk
TSC deletion23	Mouse	∼50% at 2 wk	∼1200–1500 axons at 0.5 mm†
SOCS3 deletion24	Mouse	∼40% at 2 wk	∼500 axons at 0.5 mm†	No siginificant regeneration > 1.5 mm
SOCS3 deletion + CNTF24	Mouse		∼1500 axons at 0.5 mm†	Few fibers 2–2.5 mm
PTEN/SOCS3 deletion + CNTF25	Mouse	∼60% at 2 wk	∼6000 axons at 0.5 mm†	∼20% regenerating fibers to chiasm, few to optic tract, hypothalamus, and SCN
PTEN deletion + cAMP + zymosan166	Mouse	∼36% at 10 wk	∼325 axons at 2.5 mm‡	Axons to SCN, LGN, olivary pretectal nucleus, and restitution of some visual behaviors
PTEN deletion + cAMP + zymosan165	Mouse	∼54% at 2 wk	∼800 axons at 0.5 mm†	∼1% to LGN at 6 wk

*  3 weeks postlesion.

†  2 weeks postlesion.

‡  6 weeks postlesion.

The serine-threonine kinase mammalian target of rapamycin (mTOR) was discovered as a target for the antifungal macrolide rapamycin, produced by Streptomyces hygroscopicus . ²⁶ Active during development, but downregulated in the mature CNS, and further suppressed after injury, mTOR is implicated in numerous cellular processes, and much attention has focused on the mTOR signaling pathway as a determinant of neuron survival and axon regeneration after injury. ^27,28 Genetic manipulation of signaling pathways that alter the activity of mTOR promotes some of the most significant degrees of axon regeneration achieved so far in the CNS (see Table), ^23,29,30 although restoration of sight remains an elusive goal, probably because the majority of RGC axons still fail to grow and aberrant trajectories are formed. ³¹ Identification of key growth mediators, such as mTOR, unlocks the potential for the development of new pharmacologic interventions that could be translated into therapeutic options for ophthalmologists. This article aims to explore RGC survival and axon regeneration signals, and assess mTOR-linked signaling as a potential target for pharmacological strategies to promote RGC survival and axon regeneration, and ultimately restore sight after TON. 

Neuronal Death and Axon Regeneration Failure in the Visual System

The optic nerve crush (ONC) model in adult rodents is a commonly used paradigm to study RGC responses to injury (Fig. 1). ^32,33 The ON is exposed surgically and axotomized by crushing within the meningeal sheath. The intact leptomeningeal sheath acts as a conduit, allowing approximation between the proximal (ocular side) and distal (chiasm side) axon segments at the crush site, and preserving retinal vascular perfusion through the central retinal artery. Retinal ganglion cell phenotypic markers, such as βIII-tubulin and Brn3a, are used to detect surviving RGC, while axon tracers, such as fluorogold, and axon regeneration–associated proteins, such as GAP-43, evaluate RGC axon regeneration. ^33–35  

Figure 1

 

Schematic diagram of the rodent ONC model for investigating RGC survival and axon regeneration. The RGC axons arising in the retina are crushed at the ONC site. Experimental manipulations to promote RGC survival and axon regeneration include lens injury, intravitreal delivery of putative survival/regenerative factors, implantation of a PN fragment, and targeted gene manipulation (adapted from Berry et al. 2008, ³³ with permission from IOS Press).

After axotomy in the adult rat ON, RGC axons initially sprout at the proximal lesion edge, but fail to traverse the wound to progress into the distal ON, and a glial scar develops within 8 days. ^33,36 The RGC start to die within 5 to 6 days of intraorbital ON injury, with fewer than 10% surviving by 14 days. ³⁷ However, the pattern of neuronal loss is determined by the site of the lesion along the visual pathway, as intracranial injury 8 to 9 mm behind the eye delays the onset of apoptosis to 8 days, with increased RGC survival of 80% at 4 weeks after injury, suggesting that ON glia-derived retrogradely-transported neurotrophic signals are short-term regulators of RGC viability. ³⁷  

RGC Undergo Apoptosis, but Preventing RGC Death Does Not Promote Axon Regeneration in the Mature ON

The RGC death after ON injury occurs predominantly by axotomy-induced apoptosis, associated with activated caspase signaling, ³³ which ultimately results in cleavage of intracellular proteins that maintain cellular integrity. ³⁸ The critical role of caspases in mediating injury-induced apoptosis is evidenced by enhanced RGC survival after caspase-2 inhibition by intravitreal delivery of either a modified short interfering ribonucleic acid (siRNA) ³⁹ or the pharmacologic inhibitor z-VDVAD ⁴⁰ after ONC injury. 

The Bcl family of proteins tightly regulates RGC survival. There is upregulation of proapoptotic factors and a decrease in levels of antiapoptotic Bcl-2 in the retina after ON injury, ^13,37 and overexpression of Bcl-2 in transgenic mice protects RGC from axotomy-induced cell death in postnatal ⁴¹ and adult ⁴² rodent models. Although this approach does not promote axon growth in the mature ON, ⁴³ rapid long distance RGC regeneration is seen in the ON of neonates after Bcl-2 overexpression. ⁴⁴ Overexpression of BCL-XL also is highly neuroprotective, ⁴⁵ and neutralization of proapoptotic Bax provides 95% RGC neuroprotection after ON transection. ⁴⁶ Other strategies that limit RGC apoptosis after ON injury in adults also have failed to elicit axon regenerative responses. ^33,47 Despite reports by Monnier et al. ⁴⁸ that inhibition of caspase-6 and -8, using pharmacologic caspase blockers, promotes RGC survival and axon regeneration, their demonstration of axon regeneration is less than convincing, with occasional GAP43⁺ axons detected in the distal segment of the ON. Therefore, RGC survival is a prerequisite for axon regeneration, but neuroprotection alone does not promote ON regeneration in adults in the absence of a specific axogenic stimulus. ⁴⁰  

The CNS Inhibitory Environment Limits Axon Regeneration

The classical experiments of Aguayo et al. ^14,15 demonstrated that axotomized RGC can grow axons into the permissive environment of a PN graft anastomosed to the transected proximal ON stump. However, RGC that survive axotomy do not regenerate axons spontaneously, due in part to the inhibitory milieu of the CNS. Transected RGC are exposed to growth-inhibitory ligands derived from degenerate CNS myelin and the incipient glial scar. Several myelin-derived inhibitory molecules have been characterized, including Nogo-A, myelin-associated glycoprotein (MAG), and oligodendrocyte-myelin glycoprotein (OMgp), which bind to the Nogo-66 receptor (NgR1). ¹² Ligand-receptor signaling through NgR1 requires recruitment of transmembrane coreceptors p75^NTR/TROY, LINGO-1/ AMIGO-3, ^49–53 and subsequent signaling converges on the Ras homolog gene A/rho-associated kinase (RhoA/ROCK) pathway, leading to axonal growth cone collapse. ^33,54,55 Furthermore, formation of a glial scar at the lesion site after ON transection is associated with hypertrophy and proliferation of astrocytes, and recruitment of meningeal fibroblasts, which secrete inhibitory molecules, including chondroitin sulfate proteoglycans (CPSG), ephrins, and semaphorins, which also signal Rho-A mediated growth inhibition through their cognate receptors. ^56–58 Myelin- and scar-derived inhibitors induce growth cone collapse through the LIMK/Cofilin and glycogen synthase kinase (GSK3β) pathways, respectively. ^59–62  

Strategies that overcome the inhibitory environment promote only modest axonal regeneration. ⁶³ Inactivation of RhoA/ROCK signaling blocks the effect of inhibitory ligands on RGC in vitro, and in vivo enables a small number of RGC axons to cross a lesion site and grow into the distal ON segment, an effect that is enhanced with inflammatory stimulation. ^64–68 Inhibition of NgR1 is insufficient to promote axon regeneration in the absence of an axogenic inflammatory stimulus. ^69,70 Regulated intramembranous proteolysis (RIP) of p75^NTR/TROY by neurotrophic factors ^71,72 and pharmacologic inhibition of GSK3β ^73–78 both paralyze inhibitory signaling, promoting axogenesis by protecting against growth cone collapse. Multiple extrinsic inhibitory mechanisms, therefore, combine to arrest the growth of mature CNS axons, and although blockade of inhibitory signals does enable some axon growth, additional strategies are required if functional RGC axon regeneration is to be achieved. 

RGC Survival and Axon Regeneration Are Promoted by Inflammatory Stimulation

Survival of RGC and axon regeneration can be promoted by lens injury ^19,20 ; intravitreal injection of lens crystalline proteins ⁷⁹ ; intravitreal delivery of toll-like receptor 2 agonists, such as the yeast cell wall extract zymosan ²¹ or Pam3Cys; ⁸⁰ and intravitreal implantation of a PN segment, ^19,20 with intraocular inflammation as a common feature of all these approaches ⁸¹ (Fig. 1, see Table). The regenerative state is associated with upregulation of growth-associated proteins, such as GAP-43 and SPRR1A, ⁶⁵ when RGC axon growth is possible through the normally inhibitory environment of the crushed ON. ¹² Nonetheless, the proregenerative effects are enhanced when combined with strategies to overcome inhibitory signaling, by blocking NgR1 and RhoA signaling. ^65,69  

Activated Macrophages Are Key Mediators of Inflammatory Stimulation

Influx of activated macrophages is a key event underlying the prosurvival and regenerative effects of inflammation. ^21,82 The 12 kDa macrophage-derived calcium binding protein oncomodulin (Ocm) rapidly accumulates in the inflamed eye, and is presumed to be involved in receptor binding in a cyclic AMP (cAMP)-dependent manner. ^83,84 Either an Ocm peptide antagonist or a neutralizing anti-Ocm antibody suppress inflammation-induced axon regeneration, ^70,83,84 although the role of Ocm is debated. ^85,86 Oncomodulin may act through intracellular Mst3b, a purine-sensitive protein kinase, to induce axon growth and may be linked to the MAP kinase (MAPK) signaling pathway. ^87,88  

Glial Activation and Cytokine Signaling

Intraocular inflammation also activates retinal astrocytes and Müller cells, ⁸¹ stimulating them to secrete multiple factors, including the neurotrophins nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), neurotrophin 4/5 (NT4/5), and neurotrophin 3 (NT3), as well as cytokines, such as IL-6, ciliary neurotrophic factor (CNTF), and leukemia inhibitory factor (LIF). ⁸⁹ RGC dependence on glial factors after inflammatory stimulation is implied by the failure of axon growth through an ON lesion in response to macrophage accumulation within the glial-free lesion site. ^89,90 Furthermore, vitreal and ON inflammation are RGC-protective after ON injury, but only the former is axogenic, suggesting that neuroprotective factors are transported retrogradely to RGC somata to prevent apoptosis, but induction of a growth stimulus depends on retinal glial cell activation by vitreal inflammation. ⁸⁹ Expression of CNTF and LIF is upregulated in retinal astrocytes by inflammation, ⁹¹ and the neuroprotective and axogenic effects of inflammatory stimulation are absent in mice deficient for CNTF and LIF, suggesting a pivotal role for these cytokines. ⁹² CNTF and LIF enhance neurite outgrowth in vitro, ^92–94 although the effects of intravitreal recombinant CNTF after ON injury are less robust than those induced by inflammation. ^68,94 Other neurotrophic factors supplied individually in vivo promote limited RGC survival and axon regeneration, whereas a combination of FGF-2, NT3, and BDNF act synergistically to induce robust axogenesis that is equivalent to that seen after intravitreal inflammation. ²²  

Cytokines, such as CNTF, act upon receptor complexes with a shared gp130 protein ⁹⁵ (Fig. 2), leading to association with janus kinases (JAKs), and recruitment of signal transducers and activators of transcription (STAT) proteins, particularly STAT3, and the protein tyrosine phosphatase SHP-2. ⁹⁶ Activated STAT3 translocates to the nucleus, binding to specific DNA response elements in the promoter region of target genes, ⁹⁷ and activated SHP-2 leads to signaling via the phosphoinoside-3-kinase (PI3K)/AKT pathway and MAPK/extracellular signal regulated kinase (ERK) pathways. ^12,94 The JAK/STAT system is finely regulated, and one of the genes upregulated by STAT3 activation is the suppressor of cytokine signaling 3 (socs3), a negative feedback regulator that prevents prolonged activation of this pathway (Fig. 2). Deletion of socs3 enables RGC axon regeneration after ON injury in mice, and enhances the regenerative effects of exogenous CNTF, ²⁴ whereas viral mediated overexpression of socs3 prevents axon growth. ⁹⁸ Expression of socs3 is also increased in mature axotomized RGC and, therefore, may be acting as an intrinsic brake to cytokine-mediated axon regeneration. ¹²  

Figure 2. 

 

Signaling of the mTOR pathway, with links to related signaling pathways. Neurotrophins acting upon cell membrane receptor tyrosine kinases (Trk) induces activity of the lipid kinase PI3K, which phosphorylates and activates the conversion of phosphatidylinositol (4,5) bisphosphate (PIP2) to phosphatidylinositol (3,4,5) triphosphate (PIP3). Phosphatase and tensin homolog deleted on chromosome 10 (PTEN) catalyses the reverse reaction. PIP3 recruits and activates phosphatidylinositol-dependent protein kinase 1 (PDK1), which in turn activates Akt that phosphorylates and inhibits tuberous sclerosis complex (TSC), a heterodimer comprising tuberous sclerosis protein 1 (TSC1) and TSC2. These act as a GTPase-activating proteins to stimulate the Ras homolog enriched in brain (Rheb) to upregulate mTOR activity. The mTOR forms two distinct functional complexes to influence cellular processes (see text). Akt inhibits GSK3β, which in turn disinhibits CREB-mediated NTF transcription, and disinhibits APC and CRMP2 to promote growth cone assembly. Activation of Trk stimulates the ERK/MAPK signaling pathway, which is linked to mTOR signaling. Cytokines, such as IL-6, CNTF, and LIF, activate gp130-mediated signaling through the JAK/STAT pathway to promote STAT3 activity, which is regulated negatively by SOCS3 (see text). mTORC1 also activates STAT3, linking PI3k/Akt/mTOR signaling at a downstream level. Connections between pathways enable significant cross-talk to influence multiple cellular processes.

mTOR Activity Is Linked to the Neuroprotective and Axon Regeneration Effects of Inflammatory Stimulation

Activity of mTOR has been linked to the effects of inflammatory stimulation. ⁹⁹ Intravitreal Pam3Cys, a toll-like receptor 2 agonist, prevents a decrease in RGC mTOR activity after ONC. Moreover, inhibition of mTOR activity with rapamycin attenuates inflammation-induced axon regeneration, albeit with little effect on RGC survival or short distance axon growth. mTOR is not required for the induction of a proregenerative state by inflammation, as neurite outgrowth still occurs in the presence of rapamycin in vivo and in vitro. ⁹⁹ However, at 10 days after ON transection there is reduced neurite growth from rapamycin-treated RGC after inflammatory stimulation, indicating that mTOR may have an important role in maintaining RGC regenerative ability to enable longer distance axon growth, rather than acting to initiate the process. ⁹⁹  

Developmental Downregulation of RGC Regenerative Ability

During development, axons are guided to their targets, and growth capacity of axons is reduced during the transition from embryo to adulthood to enable stabilization of synaptogenesis. ¹⁰⁰ Embryonic RGC axon growth rates are high, but decrease significantly over the later stages of development and further in adult RGC that become exquisitely sensitive to axotomy. ^101,102 Rather than merely reflecting the maturation of the inhibitory CNS environment, the development-dependent downregulation of growth ability is an intrinsic property of CNS neurons. ¹⁰³  

Cyclic Adenosine Monophosphate (cAMP)

There is a direct correlation between the age-dependent decline in neuronal cAMP and the onset of neurite growth inhibition by CNS myelin-associated factors. ¹⁰⁴ cAMP stimulates the accumulation of growth factor receptors to the cell surface, ¹⁰⁵ downregulates the activity of growth inhibitory SOCS3, ¹⁰⁶ and potentiates the effects of Ocm in mediating axon growth. ^83,84 Retinal ganglion cell exhibit a decrease in cAMP after injury, although exogenous cAMP does not increase neuronal survival in the absence of neurotrophic factors. ^107,108 Neuronal responses to cAMP depend on differential signaling through protein kinase A (PKA) and exchange protein activated by cAMP (Epac). ¹⁰⁹ Axonal growth cones demonstrate either attraction or repulsion, based on development-dependent cAMP-mediated signaling via Epac and PKA, respectively, and Epac signaling promotes axon regeneration of adult spinal cord neurons. ^110,111  

Transcription Factors

Krüppel-like factors (KLFs, a set of transcription factors) have been identified as development-dependent regulators of axonal growth. Gene expression analysis in zebrafish, which are able to mount a robust regenerative response after ON injury, found that a host of genes are upregulated in regenerating RGC, including KLF6 and KLF7, knockdown of which significantly reduced axon growth. ¹¹² KLF4 is a potent inhibitor of axon growth in embryonic RGC, and RGC lacking KLF4 show increased axon growth in vitro and after ON injury in vivo. ¹¹³ Proregenerative KLFs 6 and 7 are downregulated after birth, whereas growth inhibitory KLFs 4 and 9 are upregulated postnatally. ¹¹³ How these mediate their effects on RGC axon regenerative ability and whether there is a direct link to mTOR-related pathways is undetermined. 

Mammalian Target of Rapamycin (mTOR)

mTOR regulates protein synthesis, axonal growth and growth cone dynamics in development, ^114,115 and may be a key determinant of neuronal survival and axon growth in response to injury. mTOR activity is downregulated in CNS neurons over the course of development, and the limited residual adult mTOR activity is reduced further by axotomy. ^29,30 The mechanism of this two-step reduction in mTOR activity is not fully understood, but correlates with the reduced ability of CNS neurons to initiate the protein synthesis that is required for axon regeneration. In contrast, some injured adult PNS neurons maintain high levels of active mTOR ¹¹⁶ and their axons regenerate well, although mTOR-independent PNS regeneration also occurs. ¹¹⁷ Lack of mTOR signaling may be a major intrinsic obstacle to neuronal survival and regeneration in the adult CNS, and the importance of mTOR in inflammation-induced axon regeneration has been previously discussed. 

The PI3K/Akt/mtor Signaling Pathway

The intracellular signaling pathways mediating mTOR activation and regulation are increasingly well understood, and, therefore, manipulating mTOR to provide neuroprotection and rekindle growth makes this pathway an attractive target for potential translatable therapy for TON. mTOR activity is determined by a balance in intracellular signaling cascades (Fig. 2). The lipid kinase PI3K is activated by neurotrophins acting upon cell membrane receptor tyrosine kinases (Trk), leading to a well described canonical pathway to upregulate mTOR activity (Fig. 2). 

mTOR Forms Two Distinct Complexes With Different Substrates and Cellular Effects

mTOR exerts multiple effects on cellular function by forming two distinct functional complexes to act upon downstream substrates and influence cell survival and growth (Fig. 2). The mTOR complex 1 (mTORC1) includes mTOR and the proteins Raptor, PRAS40, Deptor, and GβL, ¹¹⁸ and is sensitive to rapamycin, which binds the FK506-binding protein FKBP12 and physically interacts with the complex to decrease its activity. ¹¹⁹ The second complex, mTORC2, also contains mTOR, Deptor, and GβL, and additionally Rictor, mSIN1, and Protor-1. ¹¹⁸ In contrast to mTORC1, mTORC2 is rapamycin-resistant although chronic exposure to rapamycin does inhibit mTORC2 assembly. ^119,120  

The two main targets of mTORC1 are ribosomal S6 kinase (S6K) and eukaryotic translation initiation factor 4E-binding protein (4E-BP1), ¹²¹ with upregulation of mTORC1 activity promoting mRNA translation and protein synthesis via these two effectors (Fig. 2). ¹¹⁵ Activation of ribosomal S6K stimulates translation of mRNA encoding proteins involved in ribosome biogenesis and protein synthesis. ^122,123 4E-BP1 is active in its hypophosphorylated form and blocks protein translation through an interaction with eukaryotic translation initiation factor 4 epsilon (eIF4E). mTORC1 acts to repress 4E-BP1 by phosphorylation, which in turn dissociates from eIF4E to enable the initiation of protein translation. ¹²⁴ Thus, mTORC1 increases the protein synthetic capacity of the cell, a key requirement in cell survival and axon regeneration. Lipid synthesis, repressed cellular degradation by autophagy, glucose metabolism, and mitochondrial function also have been linked to mTORC1 activity. ¹¹⁹ The downstream outputs and effects of mTORC2 are less well understood, although it activates calcium-dependent protein kinase Cα (PKCα), ¹²¹ to control actin dynamics, ^115,125 and activates Akt through a positive feedback pathway, thereby sustaining mTOR activation and promoting actin/microtubule assembly in axon growth cones by inhibiting GSK3β (Fig. 2). ¹²⁶ Since the integrity of the mTORC2/Akt loop is preserved after pharmacologic upregulation of mTOR downstream of Akt (e.g., by TSC1/2 inhibition), the mandatory conditions for axon regeneration of axogenic protein synthesis and growth cone mobility both will be stimulated. ¹²⁷  

Reliability of Phosphorylated S6 (pS6) as a Marker of mTOR Pathway Activity

The most widely used biochemical correlate of mTOR activity is pS6, a substrate of mTORC1. However, S6 also can be phosphorylated by mTORC1 independent pathways, such as ERK/MAPK, PKA, ^128,129 PDK1, and cell division control protein 2 (cdc2). ^130,131 Therefore, pS6 may not be highly sensitive or specific for mTORC1 activity, and reported outcomes of mTOR pathway manipulations may, in part, actually reflect effects of mTORC1/4eBP1-specific activity, mTORC2-dependent activity or signaling via the ERK/MAPK pathway (Fig. 2). The relative contribution of these pathways in RGC survival and axon regeneration is unknown. 

Other Downstream Effectors of PI3K/Akt Signaling and Links to Alternative Signaling Pathways

Glycogen Synthase Kinase (GSK3β)

mTORC1 and mTORC2 are not the only downstream effectors of the PI3K/Akt pathway. GSK3β negatively regulates growth cone microtubule assembly and is inactivated by Akt. ⁶⁰ Inhibition of GSK3β promotes actin/microtubule assembly after disinhibition of collapsing response mediator protein 2 (CRMP2) and the microtubule binding protein adenomatous polyposis coli (APC), ^132,133 and also directly phosphorylates TSC1/2 to downregulate mTOR activity ¹¹⁸ (Fig. 2). Protection against growth cone collapse and stimulation of growth cone dynamics are required for neurotrophin-induced axon growth. ¹²⁷  

The ERK/MAPK Signaling Pathway

The ERK/MAP-kinase pathway regulates protein synthesis, cell survival, and axon growth (Fig. 2), ¹³⁴ and also is activated by Trk ligands. ERK directly phosphorylates and activates mTORC1, ¹³⁵ and acts upon TSC1/2 to upregulate mTORC1 activity, ¹³⁶ and is important for the kinetics of S6 phosphorylation. ¹³⁷ ERK/MAPK and PI3K/Akt-mediated signaling may mediate distinct aspects of neuronal growth, such as axon elongation, axon caliber, and branching, ^30,103,134 effects that are enhanced by mTOR-induced axogenic protein synthesis. 

Links to gp130-Mediated Cytokine Signaling and the Wnt Pathway

CNTF/LIF signaling via the gp130 receptor complex acts through SHP2 (JAK1) to stimulate the PI3K/Akt and ERK/MAPK pathways (Fig. 2). ¹² Furthermore, mTORC1 phosphorylates STAT3 in a CNTF-dependent manner. ¹³⁸ Wnt protein signaling is associated with multiple cellular processes, including cellular development and survival, ¹¹⁸ and can increase mTORC1 activity by blocking the effect of GSK3β on the TSC1/2 complex. ^139,140 These alternative pathways highlight the opportunity for downstream cross-talk of multiple different ligand-receptor interactions, leading to many different effects on cellular processes. The complexity of intracellular signaling via multiple pathways is a significant challenge in identifying key mediators and targeting appropriate regulatory points to achieve potential therapeutic effects. 

Extrinsic Mechanisms of PI3K/Akt/mtorc1 Pathway Regulation

mTOR activity is modulated by metabolic conditions of the cellular environment, with adverse conditions, such as hypoxia, energy deprivation, and reduced nutrient availability, all inhibiting mTOR activity, with consequent restriction of protein synthesis (Fig. 3). ^141–144 The TSC1/2 complex is a key nodal point, upstream of mTORC1 and mTORC2, integrating signals from multiple sources to regulate mTOR activity and protein synthesis through mTOR. 

Figure 3. 

 

The mTOR pathway regulation by external factors. Hypoxia induces RTP801 via HIF1α-dependent and independent pathways. RTP801 promotes TSC activity to downregulate mTOR. AMPK is activated in response to impaired cellular energy production, acting directly on TSC to regulate negatively mTOR activity, and also upregulates RTP801 and directly inhibits mTORC1. Amino acid availability is signaled via Rag proteins to influence mTOR activity (see text).

Hypoxia

Hypoxia downregulates 4E-BP1 and S6K phosphorylation ^144,145 via the induction of RTP801 (Redd1/2), which acts directly on the TSC1/2 complex (Fig. 3). ^145,146 RTP801 transcription is upregulated rapidly by hypoxia inducible factor 1α (HIF1α), ¹⁴⁷ although RTP801 also is induced by hypoxia in HIF1α-deficient cells, ¹⁴⁷ suggesting that other HIF family members or HIF-independent mechanisms may have a role in hypoxia-mediated downregulation of mTOR. The hypoxic response is thought to promote ATP production in oxygen-limited conditions by promoting a shift in metabolism toward lactic acid fermentation and glycolysis. ¹¹⁹  

Energy Deprivation

Downregulation of mTOR in response to energy deprivation acts through a different mechanism (Fig. 3). The AMP-activated kinase (AMPK) is a metabolic regulator that is stimulated by impaired cellular energy production (ATP depletion or increased AMP:ATP ratio), ^145,148 and negatively regulates mTORC1 by direct phosphorylation of TSC2 ¹⁴⁸ and the raptor domain of mTORC1 (Fig. 3). ¹⁴⁹ The tumor suppressor proteins liver kinase B1 (LKB1) and p53 also act upon AMPK to inhibit mTORC1 activity further in a TSC-dependent manner. ^150,151 Moreover, AMPK itself can activate RTP801, ¹⁵² highlighting the close interaction of mTORC1 regulation in response to hypoxic and metabolic stressors. 

Nutrient Availability

The availability of amino acids also is an important determinant of mTOR function. ¹⁵³ Activation of mTORC1 by amino acids (especially leucine and arginine) is independent of TSC1/2 and relies on small GTPases, named Ras-related GTP binding proteins (Rags), ¹⁵⁴ which bind directly to the raptor component of mTORC1 to stimulate protein synthesis (Fig. 3). ^143,154 Furthermore, amino acids also may upregulate mTORC2 activity directly. ¹⁵⁵  

Mechanism of Axotomy-Induced mTOR Downregulation

The mechanism of axotomy-induced mTORC1 suppression is not fully understood. Although RTP801 mediates mTORC1 downregulation in acute hypoxia, RTP801 mRNA levels are not altered in axotomized RGC, ²³ denoting RTP801-independent mechanisms in mTORC1 suppression. AMPK/LKB1 signaling in response to metabolic stress may have a role in mediating axotomy-induced mTORC1 downregulation, although to our knowledge there are no reports of altered levels of these regulators after RGC axotomy. The tumor suppressor p53 has been linked to mTOR downregulation in response to oxidative stress, mediated by its target genes sestrin-1 and sestrin-2 activating AMPK. ¹⁵¹ The formation of stress granules (cytoplasmic RNA aggregates) in conditions of metabolic stress is reported to result in sequestration of mTORC1, whereas in homeostatic conditions stress granule dissolution occurs to promote cytosolic availability of mTORC1. ¹⁵⁶ Interestingly, glucocorticoids, such as dexamethasone, upregulate RTP801 with consequent inhibition of mTORC1 in lymphoma cells expressing the glucocorticoid receptor, ¹⁵⁷ providing a possible explanation for the failure of corticosteroids to exert a therapeutic benefit in TON. Restoration of mTOR signaling during periods of oxidative stress can prevent cell death and preserve cellular function, ^150,158 suggesting that upregulation of mTOR activity after RGC axotomy may form a therapeutic basis for neuroprotection and axogenic treatments. 

Manipulating mTOR Signaling to Promote RGC Survival and Axon Regeneration

Deletion of the pten Gene Promotes RGC Survival and Robust Axon Regeneration

There is good evidence that upregulation of mTOR signaling is a sound strategy to promote RGC survival and axon regeneration. A conditional pten gene deletion paradigm prevents RGC apoptosis in an ONC model, with approximately 45% RGC survival at 2 weeks after ONC, and also promotes ON axon regeneration with approximately 8% to 10% of the surviving RGC regenerating axons beyond the lesion site, with a small subset regenerating several millimeters as far as the optic chiasm (Fig. 4, see Table). ²³ Conditional deletion of tsc1 also promotes axon regeneration, although the effect is less pronounced than with pten gene knockdown. ²³ Similarly, deletion of either pten or tsc1/2 enhances the regenerative potential of corticospinal tract neurons, ²⁹ and DRG neurons after spinal cord and sciatic nerve lesions. ^116,117  

Figure 4. 

 

Deletion of pten promotes RGC axon regeneration. (A–D) Confocal images of optic nerves showing cholera toxin β (CTB)-labeled fibers around the lesion sites (*) at 14 days (A, C) or 28 days (B, D) after injury from PTEN^f/f mice injected with AAV-Cre (treatment, [A, B]) or AAV-GFP (control, [C, D]). Scale bar: 100 μm. Reprinted with permission from Park KK, Liu K, Hu Y, et al. Promoting axon regeneration in the adult CNS by modulation of the PTEN/mTOR pathway. Science. 2008;322:963–966. ²³ Copyright 2008 AAAS.

The conversion of PIP3 to PIP2 is blocked by pten deletion, leading to accumulation of PIP3, activation of Akt and mTORC1/mTORC2, and downstream inhibition of GSK3β, with consequent stimulation of axon protein synthesis and growth cone dynamics, respectively (Fig. 2). Using a phospho-S6 (pS6) antibody (as a marker of mTORC1 activity), the pS6 signal is abolished completely in RGC at 1, 3, and 7 days after axotomy in wild-type animals, suggesting a rapid and sustained downregulation of mTORC1 activity. ²³ By contrast, there is activation of mTORC1 in pS6⁺ RGC, correlated with axon growth after pten deletion, while rapamycin reduces pS6 signaling and diminishes the effects of pten deletion after axotomy. ²³  

mTORC1-Independent Axon Regeneration

Survival of RGC and axon regeneration after reactivation of mTOR suggests that this signaling pathway is important in the response of adult CNS neurons to injury and may be exploited usefully for therapeutic strategies to treat TON. However, some important aspects must be determined more clearly to understand fully the contribution of mTOR. Limited residual axon regeneration is seen after rapamycin delivery, ²³ indicating the contribution of mTORC1-independent mechanisms, such as GSK3β and mTORC2 activities. Consistently, a minority of regenerating RGC axons are pS6⁻, further suggesting mTORC1-independent effects. The RGC survival and axon regeneration are more marked after pten deletion compared to tsc1/2 deletion. ²³ PTEN is upstream of TSC in the mTOR signaling cascade; therefore, the differential effects of pten and tsc1/2 deletion after ONC are likely to reflect activity upstream of TSC, for example through GSK3β-mediated effects (Fig. 2). 

Discrepancy Between RGC Survival and Axon Regeneration

Deletion of pten protects approximately 45% RGC from apoptosis, although axon regeneration is limited to fewer than 10% of RGC surviving axotomy. This discrepancy demonstrates that axon regeneration is not an inevitable outcome of survival, but reflects a balance between direct mTOR-mediated effects on protein synthesis, and links to other signaling pathways that may have a more profound influence on regulation of apoptosis than axon growth. Activation of S6K, an output measure of mTOR upregulation, promotes the phosphorylation of the Bcl2-associated death promoter (BAD) in brain astrocytes to limit ischemia-induced apoptosis. ¹⁵⁹ Conversely, PTEN upregulation by oxidative stress, which counteracts mTOR activity, is linked to the induction of apoptosis ¹⁶⁰ and pten deletion prevents cortical neurons undergoing apoptosis after hypoxia-ischemia injury. ¹⁶¹ The mTOR signaling, therefore, is closely linked to apoptosis/cell death pathways, and its activity may be important for synthesizing the raw materials required for cell survival and axon growth, and the maintenance of a regenerative ability, rather than initiating long-distance axon growth in the absence of other signaling. Thus, a coincident growth promoting stimulus also may be required in addition to mTOR upregulation as a viable therapeutic strategy in TON. 

There is some evidence from CNS diseases that clinical deficit may arise in part from secondary degeneration of initially undamaged neurons, ¹⁶² and secondary RGC death has been reported after incomplete ON injury by partial ONC. ^163,164 The mechanisms of secondary neuronal degeneration may include accumulation of cellular byproducts from locally degenerating neurons/glia, loss of local trophic support, and disturbed local physiologic and immunologic equilibria. ¹⁶⁴ Evidence for mTOR signaling in RGC survival and axon regeneration has come from studies of complete axonal transection by ONC, but it is feasible that mTOR-induced NTF production mediating RGC survival may have a beneficial role in limiting secondary neuronal degeneration after incomplete ON injury. 

Combinatorial Approaches to Promote RGC Survival and Axon Regeneration

Given the multiple mechanisms acting in concert to regulate neuronal survival and axonal regrowth, it may be no surprise that a single targeted manipulation is insufficient to protect all RGC from apoptosis and stimulate axon regrowth after ON injury, and combinatorial approaches consistently yield stronger survival and axon regenerative responses than do individual manipulations (see Table). Inflammatory stimulation combined with pten deletion promotes long-distance regeneration over 5 mm, and some target reinnervation, ¹⁶⁵ and may be sufficient to regain partially some visually-dependent behavior. ¹⁶⁶ Combined deletion of pten and socs3, together with CNTF administration, promotes a 10-fold increase in the numbers of regenerating axons at 2 mm beyond the ONC site compared to deletion of either gene alone. ²⁵  

Taken together, these results support the use of combinatorial approaches to prevent apoptosis, stimulate transformation into a regenerative phenotype, and maintain the growth state for long-distance RGC axon growth. Upregulation of PI3K/Akt-pathway signaling activates mTOR, to promote cell survival and protein synthesis, which determine regenerative competence and maintenance of a regenerative state, and also triggers growth cone advance by suppressing GSK3β activity. Cytokine signaling through the CNTF-gp130 pathway also induces an axogenic programme (see earlier). Combining activation of mTOR signaling by pten deletion with either an inflammatory stimulus ^165,166 or with activation of the JAK/STAT pathway ²⁵ enables significant axonal regeneration, but neither approach has required specific neutralization of external axon growth inhibitory molecules, probably because protection against growth cone collapse is mediated through neurotrophin-induced RIP and by inhibition of GSK3β through Akt phosphorylation. ^{74–77,167–169} Also, mTORC1 may affect the sensitivity of RGC to myelin-associated inhibitory molecules, ^99,170 blinding growth cones to regeneration-arresting signals, possibly by upregulating neurotrophic factors, which induce RIP of the p75^NTR complex. ⁷¹ In all ON regeneration paradigms, scar tissue fails to develop at the lesion site, suggesting that scarring represents a failure of axon regeneration, and that targeted strategies to overcome scarring and inhibition may be redundant if robust axon growth can be initiated. ^33,171  

Future Challenges: Exploiting mTOR for Clinically Relevant RGC Survival and Regeneration

Unlocking Axon Regeneration in Nonresponding RGC

Despite the identification of molecular mechanisms surrounding axon regeneration failure in the mature CNS, and promising experimental approaches to overcome these to induce RGC survival and axon regeneration after ON injury, a number of challenges remain. At present, only a small proportion of RGC have been stimulated to regenerate axons, despite manipulation of many candidate signaling pathways, including upregulation of mTOR activity. Understanding why the majority of axons do not regenerate after these axogenic treatments may hold the key to visual recovery in patients blinded by TON. While PI3K/Akt activation of mTOR improves RGC survival and contributes to axon regeneration, upregulation of other pathways, such as MAPK, GSK3β, and CNTF/LIF-mediated JAK-STAT signaling, together with increasing cAMP may enable more robust axon regenerative responses. Moreover, a deeper understanding of the role of glia, and more subtle control mechanisms of cellular signaling also is required. Temporal dynamics of pathway activation may be relevant, as the duration of receptor signaling mediates specific effects in response to activation by different ligands. ¹⁷² Epigenetic changes ¹⁷³ and micro-RNA modulation of gene expression, including those related to mTOR signaling, have pivotal roles in neuronal survival and axon growth, and require further investigation. ^174,175  

Forming the Right Connections

Even if the goal of ON axon regeneration is realized, RGC axons may not be guided appropriately to their central targets, ³¹ aberrant functional synapses may be formed, and topographic representations mismatched. ¹³ Anatomically conserved functional synaptic connections are possible after nerve regeneration in hippocampal neurons, ^176–178 and target-specific axon regeneration has been reported in the murine visual system using a combination of pten deletion, zymosan, and a cAMP analogue. ¹⁶⁶ However, utilizing the same combinatorial treatment strategy, recent imaging studies using light sheet fluorescence microscopy for 3-dimensional evaluation of regenerating axons could not detect the same degree of central reinnervation, and reported significant aberrant axon growth trajectories. ³¹ The reason for the discrepancy in these findings is unclear, although aberrant growth of regenerating axons in the ON has been reported previously after intravitreal PN implant as the growth stimulus. ¹⁸ Functional plasticity of the CNS may overcome deficiencies in anatomic connections. ¹⁷⁹  

Translating Molecular Biology to the Ophthalmic Clinic

Once the optimal biochemical conditions for RGC survival and axon regeneration are elucidated, a further challenge remains in translating them into a clinical approach. Conditional gene deletion is not translatable and gene therapy is in its infancy, but siRNA-based therapies to knock-down target gene function may offer the potential for more rapid delivery to patients. ¹⁸⁰ Clinical trials of intravitreal siRNA treatments, including those targeting RTP801, already have been reported for ocular disorders, such as diabetic macular edema and age-related macular degeneration. ^181–183  

Potential Adverse Effects of Upregulating mTOR

Despite the enthusiasm for upregulating mTOR activity as a component of a therapeutic approach for TON, a note of caution should be sounded. Hyperactivation of the PI3K/AKT/mTOR and JAK/STAT pathways is found in many malignant diseases. ^184,185 Sustained mTOR overaction is associated with aberrant protein translation and nerve growth, linked to epilepsy, adverse inflammatory responses in traumatic brain injury, and intellectual disorders, such as autism and cognitive decline in dementia. ^27,186 Achieving the balance of regenerative growth promotion without inducing malignant proliferation or dysregulated sprouting remains a formidable challenge. Therefore, development of treatments based upon modulation of the mTOR pathway must be mindful of potential adverse and off-target effects, although tissue-specific drug delivery and short-term signal upregulation may protect against possible detrimental outcomes. 

Conclusions

With no clinically effective treatments for TON, patients are at risk of lifelong visual impairment. Laboratory studies are unlocking multiple extrinsic and intrinsic factors, and the candidate signaling pathways responsible for RGC death and axon regeneration failure in the adult visual system that could be targeted for clinical treatment. The PI3K/Akt pathway, which mediates axon growth and protein synthesis through GSK3β and mTOR signaling, respectively, is a promising candidate pathway. Altering the balance of signaling in mTOR and linked pathways to promote RGC survival and axon regeneration, and translating such laboratory studies into the clinic remain significant challenges, although increasing evidence suggests that the task may not be insurmountable. 

Acknowledgments

Supported by Medical Research Council Grant MR/J011584/1. 

Disclosure: P.J. Morgan-Warren, None; M. Berry, None; Z. Ahmed, None; R.A.H. Scott, None; A. Logan, None 

Footnotes

 RAHS and AL are joint senior authors.