Glial Cell Line-Derived Neurotrophic Factor and Focal Ischemic Stroke

Abstract

Focal ischemic stroke (FIS) is a leading cause of human debilitation and death. Following the onset of a FIS, the brain experiences a series of spatiotemporal changes which are exemplified in different pathological processes. One prominent feature of FIS is development of reactive astrogliosis and glial scar formation in the peri-infarct region (PIR). During the subacute phase, astrocytes in PIR are activated, referred to as reactive astrocytes (RAs), exhibit changes in morphology (hypotrophy), show increased proliferation capacity, and altered gene expression, a phenomenon known as reactive astrogliosis. Subsequently, the morphology of RAs remains stable, and proliferation starts to decline together with the formation of glial scars. Reactive astrogliosis and glial scar formation eventually cause substantial tissue remodeling and changes in permanent structure around the PIR. Glial cell line-derived neurotrophic factor (GDNF) was originally isolated from a rat glioma cell-line and regarded as a potent survival neurotrophic factor. Under normal conditions, GDNF is expressed in neurons but is upregulated in RAs after FIS. This review briefly describes properties of GDNF, its receptor-mediated signaling pathways, as well as recent studies regarding the role of RAs-derived GDNF in neuronal protection and brain recovery. These results provide evidence suggesting an important role of GDNF in intrinsic brain repair and recovery after FIS, and thus targeting GDNF in RAs may be effective for stroke therapy.

1. Focal Ischemic Stroke and Reactive Astrocytes

Cerebral focal ischemic stroke (FIS) is a leading neurological disorder associated with neuronal death and brain damage, disabilities leading to a major impact on public health. FIS is induced by development of a thrombus in the arteries supplying blood to the brain, which is a complex process leading to changes in microenvironment in the Central Nervous System (CNS). Patients with FIS may suddenly experience paralysis, impairment of speech, and loss of vision. Most survivors develop disabilities which are often permanent. Despite of intense basic and translational research being taken towards FIS, treatment options are still limited. Currently, tissue plasminogen activator (tPA) is the only FDA approved drug which has limited application with a window of 3-5 hours after the onset of FIS. Consequently, improvement of stroke outcomes and recovery of survivors frequently depend on self-brain repair and rehabilitation. Recent advances on mechanisms of post-stroke recovery include improving growth factor release, intrinsic antioxidant capacity, and enhanced neurogenesis and angiogenesis (1-4).

Astrocytes are a major and the most diverse glial cell type in CNS. They are highly heterogeneous in morphology and function and demonstrate remarkable adaptive plasticity that defines the functional maintenance of the CNS during development and aging, and under pathological conditions (5). With their intimate contact with neurons, these cells fulfill many functions including maintenance of ion homeostasis, storage and distribution of energy substrates, and controlling the development of neural cells and synaptogenesis (6-10). Brain injury and other chronical neurological disorders cause the activation of astrocytes, a phenomenon generally known as reactive astrogliosis. The activated astrocytes is called reactive astrocytes (RAs). Reactive astrogliosis is observed in most if not all of the brain disorders. The commons molecular and morphological features of RAs include hypertrophy of astrocyte processes and upregulation of glial fibrillary acidic protein (GFAP), the key constituent of astrocyte intermediate filaments. However, the properties of RAs are disease context dependent and also depend on disease stages and brain regions (e.g., brain injury vs neurodegenerative diseases, mild vs severe injury, and acute vs chronic stage). For comprehensive review of reactive astrogliosis and RAs, readers can consult many nice review articles (5, 11, 12).

The pathological process of FIS has unique features compared with neurodegenerative diseases due to its focal decrease of blood flow in the brain. Following the onset of a FIS, the brain experiences a series of spatiotemporal changes associated with different pathological processes. In the ischemic core, neurons die rapidly after the onset of FIS due mainly to the impairment of energy production and breakdown of ion homeostasis. In the peri-infarct region (PIR) (also commonly called penumbra), brain tissue is hypo-perfused with collateral blood flow as an attempt to preserve partial energy metabolism (Figure 1). Accordingly, the primary goal for stroke therapy is to salvage the PIR tissue after FIS. During the subacute phase (days after a FIS), surviving neurons in the PIR exhibit active structural and functional changes in synaptic wiring and mapping, and these changes may contribute to long-term neural regeneration and functional recovery (13-17). Another hallmark of FIS is glial scar formation in the PIR. During the subacute phase, RAs in the PIR exhibit spatiotemporal changes in morphology (hypotrophy) and proliferation capacity, resulting from altered metabolic functions and gene expression (18-20). Under this condition, RAs become hypertrophic with large processes that can be revealed by large upregulation of GFAP (Figure 1). Upon prolonged time (in chronic phase) following FIS, proliferation of RAs starts to decline and glial scars are stabilized (18-23). Reactive astrogliosis and glial scar formation eventually cause substantial tissue remodeling and structural changes in the PIR. Under normal conditions, astrocytes are quiescent in term of proliferation, but after FIS, RAs have high capacity to proliferate and reaching a peak around 3-4 days (18, 19). The highly active RAs on 3-4 days post-stroke are needed to meet the high energetic and biosynthetic demands for proliferation. RAs have been demonstrated to offer protective effects after FIS and injury (24-27). Proliferating RAs may impact tissue preservation, repair/remodeling, and functional outcome. Specific deletion of proliferating RAs after brain injury has been shown to prevent repair of the blood–brain barrier, increase immune cell infiltration as well as neuronal degeneration(28). The molecular phenotype exhibiting rapid and transient induction of gene expression in RAs after ischemia supports the notion that RAs may offer beneficial or protective effects to neurons (29).

Figure 1. Brain damage and reactive astrogliosis after FIS.

A) Nissl staining showing brain damage in the mouse cortex after PT. B) Double staining of Brdu and GFAP in the cortices of a control (left) and an ischemic mouse 4 days after PT (right). IC-Ischemic core; PIR-Peri-infarct region. Notice that GFAP is highly upregulated in RAs, which is a hallmark of FIS. Adapted from reference Li H. eat al. (16).

Since neuron-centric strategies have not achieved major breakthroughs in stroke therapy, astrocytes have been indicated as a promising target for neural repair in stroke and other brain diseases through either cell or non-cell autonomous effects on neuronal survival and regeneration (9, 20, 21, 24, 30-38). The potential beneficial effects of RAs on brain recovery have been discussed in many reviews (20, 21, 24, 39, 40); however, molecular mechanisms underlying their protective effects remain elusive. Recent studies have suggested that a potential mechanism for beneficial effects is the ability for RAs to release different soluble factors including BDNF, NDF, GDNF, CNTF and CNDF that are important for brain to recover (24, 41-43). This article is intended to focus on GDNF and its effects on brain protection and recovery after FIS.

2. GDNF Ligand Family, Receptor Systems, and Signaling

2.1. GDNF function in the CNS

GDNF is a member of transforming growth factor (TGF)-β super family; it is also a potent neurotrophic factor that promotes neuronal survival, neural progenitor differentiation, and synapse formation (44-49). GDNF was originally isolated from the supernatant of a rat glioma cell-line (44). Initially, it was thought to act as a trophic factor for embryonic midbrain dopaminergic neurons, but was later found to have pronounced effects on other neuronal subpopulations (50). GDNF family proteins were first purified and proved to promote the survival of midbrain dopaminergic neurons. Later, there is more evidence showing that not only dopaminergic neurons, but many other cell types can also be targeted by GDNF. Moreover, the role of GDNF can be changed in different stages. For example, GDNF and its receptors exert a non-survival function during neuronal development (51) and favors different types of neuronal precursors by promoting axonal growth. After differentiation, GDNF functions as a survival factor (52), through regulating the expression of key enzymes to promote survival of neurons and mediate dendritic or electrical maturation, particularly those related to dopaminergic phenotype. GDNF can also promote neurogenesis, neuronal maturation and mediate neuromuscular junction function in motor neuron system (53). Moreover, the ability of GDNF to modulate proliferation of neuroblastoma has been shown to closely associate with conditions such as depression or chronic pain (52).

2.2. The GDNF protein

GDNF and GDNF-family ligands are comprised of neurturin, artemin and persephin, and are thought to belong to the TGF-β superfamily (47). In their structure, all have cysteine knot motif and three disulphide bonds, and they usually contain seven cysteine residues with same relative spacing. This kind of proteins normally function as homodimers (Figure 2).

Figure 2. GDNF family ligands and their receptors.

The four members of GDNF family ligands include GDNF, neurturin, artemin, and persephin. They are homodimers and bind with high affinity to one of the four members of the GDNF receptor α family (GFRα1-4). These receptor-ligand complexes can interact with and activate the canonical GDNF receptor ‘rearranged during transfection’ (RET), which is a receptor tyrosine kinase (TK). GDNF can also activate alternative GDNF receptors, such as the neuronal cell adhesion molecule (NCAM). The intracellular domain of RET can be phosphorylated and ubiquitinylated. Adapted from Edgar R. Kramer and Birgit Liss (155).

Because all GDNF family proteins are glycosylated, intracellular modification is needed for maturation of these proteins. Similar to other secreted proteins, GDNF is produced in the form of a precursor, which is then cleaved and activated upon secretion. The mature form of GDNF has 134 amino acids, with a molecular weight around 30-42 KD. The N-terminal consists of a heparin-binding domain which can bind with heparan-sulphate extracellular proteoglygan (54). During the past decades, synthesis and modification of GDNF have been investigated extensively. However, knowledge on specific enzyme(s) that cleaves and activates GDNF protein is still limited and thus needing future investigation.

C6 glioma cells is considered as astrocyte-like cell (55). GDNF can be highly synthesized and released from these cells. In fact, glioma C6 cells are widely used as a model to investigate the regulatory mechanisms of GDNF synthesis. Many studies on GDNF release measure GDNF level in the culture medium. Hisaoka et al. demonstrated that the antidepressant, amitriptyline, could increase GDNF mRNA expression and GDNF release, and may be regulated by the MEK/MAPK (mitogen-activated protein kinase) pathway (56). However, phosphoinositositide-3-kinase (PI-3K)/AKT pathway was shown to play a positive role in FGF-2-stimulated GDNF release in C6 glioma cells , but this release was independent of the p44/p42 MAPK or stress-activated protein kinase/c-Jun N-terminal kinase (SAPK/JNK) (57). Verity et al. demonstrated that diverse biological factors were capable of modulating the levels of GDNF and that multiple signal transduction systems could regulate GDNF synthesis and/or release from rat C6 glioblastoma (58). Proinflammatory cytokines such as interleukin-1β, interleukin-6, tumor necrosis factor-α, and lipopolysaccharides could largely elevate GDNF release. Activation of protein kinase C (PKC) and Ca²⁺ ionophore can also enhance GDNF release. Study by Tanabe et al. also demonstrated that IL-1β could stimulate GDNF release through the pathways of IκB-nuclear factor kappa B (NF-κB), p38 MAPK, p44/p42 MAPK and JAK-STAT3 (59).

GDNF can be released by astrocytes or other types of gilal cells to support the survival and function of neurons, especially under injured conditions. Kuno et al. examined the regulatory mechanisms of GDNF production induced by tumor necrosis factor-α (TNF-α) in cultured astrocytes(60). Their study showed that cultured astrocytes expressed both TNF-α receptor 1 (TNFR1) and TNFR2, and that activation of these receptors by TNF-α could promote expression of both NGF and GDNF. In addition, not only exogenous TNF-α but also TNF-α produced by astrocytes could induce NGF and GDNF production in astrocytes. Their studies suggest that TNF-α can contribute to the production of neurotrophic factors in response to inflammation.

While the function of GDNF has been studied extensively, less is known about the mechanism(s) underlying its release in physiological and pathological conditions. Lonka-Nevalaita et al. demonstrated that KCl-induced depolarization to cultured neurons could increase the secretion of β precursor of GDNF and its corresponding mature GDNF in a Ca²⁺-dependent manner (61). Using neuron-like PC-6.3 cells, these investigators further showed that both β precursor GDNF and mature GDNF were localized primarily in secretogranin II and Rab3A-positive vesicles of the regulated secretory pathway. However, it is not clear whether GDNF is released from astrocytes through similar pathways, since astrocytes are non-excitable cells. Therefore, different secretary pathways may depend on cell type and stimulation.

2.3. GDNF receptors

RET is the most common signaling receptors for GDNF, and the RET gene is a proto-oncogene which can be activated by a DNA rearrangement mechanism (47, 62, 63) (Figure 2). This gene is primarily expressed in the peripheral enteric, sympathetic and sensory neurons, as well as in motor and dopamine neurons (63). RET receptor is a tyrosine kinase (TK) with four cadherin-like repeats and a typical TK domain (47). However, RET can be activated when GDNF firstly binds with a novel class of protein called GDNF family receptors (GFR)-α receptors located in the plasma membrane through the glycosylphosphatidyl inositol (GPI)-anchored protein (64, 65). Currently, four different GFRα receptors have been identified: GFRα1, GFRα2, GFRα3 and GFRα4, and these can bind with GDNF, neurturin (NRTN), artemin (ARTN) and persephin (PSPN), respectively(64). GDNF family proteins first form a complex with one of the four GFRα proteins, which trigger aggregation of two RET molecules, and subsequently result in transphosphorylation of specific tyrosine residues in the TK domain which then initiate intracellular pathways (47, 64, 65). In addition to GFRα, effective RET phosphorylation needs the presence of heparan sulphate, which can facilitate the interaction between GDNF and RET (54, 66). This phenomenon can also be observed upon using a high concentration of GDNF in the absence of heparan sulphate (66).

In recent years, there is evidence that other binding mechanisms or receptors can also mediate GDNF activity. Interestingly, since GDNF mutants deficient in GFRα1 binding activity could still activate RET (67), this result suggests that RET activation is independent of GFRα. Apart from RET, GDNF can trigger intracellular pathways through other receptors. In tissues like forebrain or inner ear, RET is not present, despite of existence of GFRα, thus suggesting that other transmembrane proteins can mediate GDNF activity (68). Neural cell adhesion protein (NCAM), has been proved as a transmembrane protein serving as a second GDNF receptor (69). GDNF binding to NCAM can facilitate Schwann cell migration and axonal growth in hippocampus and cortex (64). Although new activating mechanisms and receptors have been identified, our knowledge is still limited and future studies are highly warranted.

2.4. The GDNF signaling pathways

RET-dependent signaling pathways.

After binding with GDNF-GFRα complex, RET can activate many downstream signaling pathways inside and outside the lipid raft, a membrane domain essential for cell survival, differentiation, proliferation, neurite outgrowth and synaptic plasticity. Lipid rafts present on the cell surface are essential signaling compartments for cell adhesion, axon guidance, and are crucial for signal transduction (70, 71). Lipid rafts consist of dynamic assemblies of cholesterol and sphingolipids on the outer leaflet, and phospholipids in the inner leaflet (47, 70, 71). Lipid rafts also contain interactive proteins that can change in size or composition in response to stimulation of signaling pathways (47, 70, 71). Many proteins such as the GPI-anchored proteins show high affinity for rafts. Signal transduction of RET-dependent pathways appear to rely on the colocalization of RET and GFRα in the lipid rafts (72). GFRα, by virtue of GPI anchors, is normally located in lipid rafts, whereas RET is usually outside of rafts (73, 74). GDNF binding with GFRα could recruit RET to the lipid rafts, which is essential for activating downstream pathways (47, 73, 74).

Once RET binds with GFRα-GDNF complex, the tyrosine residues of RET can be phosphorylated and become activated (75). Activated RET can serve as binding sites for various intracellular proteins within the cells (76, 77). Tyr1062 can bind at least five different proteins: Shc (Src homologous and collagen-like proteins), FRS2 (fibroblast growth factor receptor substrate 2), DOk4/5 (downstream of TK 4/5), IRS1/2 (insulin receptor substrate 1/2) and enigma (47, 76, 77). These proteins have many downstream targets. In addition, activated RET can also activate PI-3K and MAPK and their downstream pathways (78-80). These pathways are known to contribute to neuronal differentiation and neurogenesis. In recent years, there is evidence that activated RET can also activate the phospholipase C-γ (PLC-γ) pathway, and thus enhance neurotransmission together with the PI-3K pathway (47, 81).

Interestingly, phosphorylation of RET can take place in different tyrosine residues after binding with the GDNF-GRFα complex. Four key residues have been identified: Tyr905, Tyr1015, Tyr1062 and Tyr1096 (64). Phosphorylation of these four residues led to stimulate a similar profile of downstream signaling pathways MAPKs ERK1 and ERK2 and the AKT kinase (75). Although different GDNF-GFRα complexes can trigger phosphorylation at different sites, it appears that different isoforms of RET can initiate specific signaling pathway (47, 75, 82). There are two isoforms of RET receptors, namely, RET9 and RET51, which differ in carboxyl terminal (72). These two isoforms do not associate with each other, and their signaling pathways are quite different. As compared with RET9, the longer one, RET51, interacts more strongly with ubiquitin ligase Cb1(83). RET51 can also interact with CrK1 while RET9 cannot. Mice lacking the long isoform appear normal in RET null mutation; however, lacking the shorter form led to abnormal kidney and enteric aganglionosis (84). These abnormalities could be rescued by RET9 isoform, suggesting that the short form is important in RET signaling in embryonic development.

RET independent signaling pathways

GDNF can activate intracellular signaling independent of the RET receptor (47). In a RET deficient cell line and primary neurons, GDNF-GFRα complexes can activate another protein called SFK (72, 85). Activated SFK was shown to phosphorylate a series of proteins such as MAPK, PCL-γ, CREB (cAMP response element binding proteins) and FOS. GDNF can promote biochemical and biological responses in conditionally immortalized neuronal precursors that express high levels of GFRαs but not RET (85). GDNF treatment did not activate the Ras/ERK pathway in these RET deficient cells, but stimulated a GFRα1-associated Src-like kinase activity in the detergent-insoluble membrane compartments, which induced rapid phosphorylation of cAMP response element-binding protein, up-regulation of c-fos mRNA, and cell survival. GDNF can also initiate signal transduction through other receptors such as NACM (69). In epithelial cells, GDNF-GFRα complexes can interact with NCAM receptor leading to activation of Fyn, a Src like protein. Fyn can in turn activate MET and initiate downstream signaling pathways. However, how Fyn is activated or how activated MET can initiate the signaling pathway is not fully understood. The RET independent pathway presents novel signaling mechanisms directly or indirectly mediated by GFRα receptors acting on a cell-autonomous manner, and thus uncovering an unexpected intersection between short- and long-range mechanisms of intercellular communication. The mechanism of RET independent signal transduction needs future investigation.

In sum, there is strong evidence suggesting that GDNF can act through a variety of pathways to mediate development and differentiation, and thus offering protection for impaired neurons after FIS.

3. GDNF in Astrocytes and Neuronal Protection after FIS

3.1. Expression of GDNF and its receptor in RAs

The ability for GDNF to protect dopaminergic and spinal motor neurons has led to consideration for this compound as a drug candidate for the treatment of Parkison’s Disease (PD) as well as other neurodegenerative diseases (86, 87). GDNF could be released from cultured astrocytes (88-91), possible because cultured astrocytes are more akin to RAs (92). GDNF is secreted to extracellular space as a glycosylated mature protein and can mediate cellular responses through GFRs (86). For in vivo studies based on mRNA analysis, GDNF was expressed in the striatum and skeletal muscle (62), targeted fields for dopaminergic substantia neurons and motor neurons, respectively. In normal brain, GDNF is mostly confined to neurons and is absent from or expressed at low levels in astrocytes (93-95). Consistently, RNA-seq study also showed that mature mouse neurons express higher GDNF mRNA than mature mouse astrocytes (96). However, many studies have demonstrated upregulation of GDNF in RAs after FIS (93, 97-100). After stroke, a 3-fold increased expression of GDNF was mainly limited to RAs (100, 101). In situ hybridization study also indicated elevation of mRNA of GDNF and the increased number of GDNF mRNA-expressing cells in the cortex and striatum after middle cerebral artery occlusion (MCAO), whereas no changes in mRNA expression for neurturin or persephin were detected (100). At 3-7 days after MCAO, both RAs and surviving neurons showed increase in GDNF (93). In the PIR of cerebral cortex and striatum, expression of GDNF mRNA and protein increased as early as 2 h after ischemia-reperfusion, and this was followed by a decline and another increase at 72 h (99). Double staining showed that the earlier peak of GDNF expression was of neuronal origin and the later peak of glial origin. Using Western blot analysis, Zhang et al. showed that GDNF levels increased for a sustained period after photothrombosis (PT)-induced FIS, and immunostaining of GDNF and GFAP indicated significantly increase in RAs (102).

Besides expression of GFRs in cultured cortical astrocytes and neurons (103), the expression of receptors for GDNF family proteins were also increased after stroke (100, 101). After a 2-h MCAO, both c-Ret, GFRα1 and GFRα2 mRNA levels were markedly increased outside the ischemic lesion in the ipsilateral cortex at 6–24 h, whereas GFRα2 expression was decreased in the cortical areas both within and outside the lesion (100). In a rat model of occlusion of the right carotid arteries, both RET and GFRα mRNA level were markedly increased in the cortex, striatum and hippocampus 6-24 hours after ischemia (100). Astrocytes in rats treated with quinolinic acid or kainic acid also showed increased expression of GDNF as well as its receptor GFRα1 (101, 104). Similarly, after mechanical injury of spinal cord in adult rats, GDNF and GFRα1 were also upregulated in RAs (105). Upregulation of GDNF and its receptor signaling pathways in RAs after stroke strongly suggests that RAs may serve as an important source of GDNF to enhance neuroprotective and neuroregenerative responses during the post-ischemic time.

3.2. The role and mechanisms of astrocyte-derived GDNF in stroke outcomes

The brain tissue is known to have a remarkable capacity for spontaneous recovery after FIS, although such mechanism(s) remains largely unclear. Substantial interest has focused on the potential protective effects of GDNF owning to its upregulated production as well as expression of its receptors RET and GFRα-1 (93, 95, 97-100). In fact, based on treatment using recombinant GDNF and molecular genetic methods, multiple mechanisms for GDNF to protect neurons and brain after stroke have been investigated (Figure 3). For example, recombinant GDNF was shown to offer protection after MCAO (106-111). GDNF could reduce apoptosis via the caspase-3 dependent pathway (112), through upregulation of anti-apoptotic Bcl-2 and Bcl-Xl (108, 113), and potentiate and prolong activation of the p-AKT pathway (109). After permanent MCAO, GDNF treatment in cerebral cortex could significantly reduce the immunoreactivity of caspases-1 and caspases-3 in TUNEL positive apoptotic cells, suggesting that GDNF could inhibit DNA fragmentation in apoptotic pathways (108). GDNF could also inhibit excitotoxicity after stroke (114). In an in vivo excitotoxicity model induced by administration of a unilateral dose of N-methyl-d-aspartate (NMDA) to the hippocampus (a mechanism that could induce lesion in the CA1 region), prior transduction of a lentiviral vector expressing GDNF showed protection against neuronal loss and significantly reduced lesion size, thus protecting the hippocampus from excitotoxic damage. Administration of GDNF in the injured cortex could also reduced NMDA-induced calcium influx (103). In hippocampal brain slices, pretreatment with GDNF was shown to significantly attenuate the NMDA-induced cell death (106). The protective response was related to the increased expression of GDNF after administration of NMDA (115). There is also evidence showing that GDNF could attenuate slowing of NMDA-induced excitotoxic neuronal death through activation of the MAPK pathway (103).

Figure 3. Potential mechanisms of RAs-derived GDNP on neuronal protection after stroke.

Astrocytes under normal condition express and release little GDNF. After a FIS, GDNF is upregulated and released from reactive astrocytes (RAs), which can inhibit excitotoxicity, apoptosis, oxidative stress, and modulate synaptic loss of neurons, thereby promoting neuronal survival and brain repair, and improving long-term stroke outcome. Red dots represent GDNF molecules.

Increased oxidative stress is known to contribute to neuronal death after stroke, and thus affecting brain recovery. Along this line, recombinant GDNF was shown to reduce oxidative stress and protect neurons via transcriptional regulation of glutathione synthesis (116). Increased oxidative stress can be attributed to two important oxidative species, namely, the ROS (reactive oxide species) and RNS (reactive nitrogen species) (1, 117). Under ischemic conditions, impaired ability of neurons to maintain normal ionic gradients can lead to excessive influx of calcium, and subsequently activating nitric oxide synthase and NADPH oxidase to produce more RNS and ROS (1, 118). After stroke, GDNF was shown to prevent degeneration of motor neurons through inhibiting the production of RNS and ROS (119). Indeed, treatment of GDNF after stroke could inhibit nitric oxide synthase (NOS) and thus decreased production of RNS (120). GDNF could also downregulate ROS and prevent the over influx of calcium (103).

GSH is known to serve as an antioxidant within cells to defend against oxidative stress-induced injury. Activation of the pentose phosphate pathway (PPP) is considered as the first line of defense response against oxidative stress-induced cell death (121, 122). The production of GSH in this pathway is controlled by NADPH which is mainly generated through the PPP. G6PD is the rate-limiting enzyme in the oxidative stage of PPP and is of central importance to provide NADPH. G6PD converts G6P to 6-p-gluconolactone and generates NADPH, which is required for regeneration of GSH in antioxidant pathways to reduce oxidative stress (123, 124). GDNF treatment can increase the level of reduced glutathione (GSH) (125), suggesting that GDNF not only can directly reduce oxidative species but also increase the antioxidative capability of cells and tissues.

Studies using recombinant GDNF have demonstrated that the neuronal and brain protective effects of this factor can be a promising candidate for stroke therapy. However, there are few weaknesses exist in term of mechanism. First, local delivery of recombinant GDNF protein can only affect a small region of tissue surrounding the injection site. Secondly, most studies have focused on evaluating the effects of GDNF on brain infarction and cell death in the acute phase, but whether these effects prevail upon long-term neural regeneration and motor functional recovery have not been studied in detail. Thirdly, GDNF administration is not cell-type specific, therefore, leaving the respective roles of neuron- or astrocyte-derived GDNF in brain protection and recovery unclear. Lastly, some toxic effects were found upon continuous administration of recombinant GDNF to the primate PD model (86).

There is increasing evidence indicating a non-cell autonomous effect of astrocytes or RAs on neurodegeneration and death under different pathological conditions (35, 126). Genetic strategies such as viral transduction to increase local production of GDNF have been tested for stroke therapy (127-131). However, few studies have used molecular genetic approach to test whether and how astrocyte specific GDNF could provide neuronal and brain protection after FIS. With the availability of genetic tools such as astrocyte-specific driver transgenic mouse lines and astrocyte-specific adeno-associated virus (AAV) for gene deletion and overexpression (132, 133), dissection of exclusive roles of astrocytes in neuronal circuit and in diseases associated with neurodegeneration is becoming feasible. For example, GLAST-CreER^T2 mediated recombination has been used to express fluorescent reporters to study the morphology and Ca²⁺ signaling in astrocytes, and to label adult-born granule cells (134-136). To study the effects of RA-derived GDNF on brain protection and recovery, Zhang et al. generated astrocyte-specific, inducible and conditional GDNF knockout (cKO) mice, i.e., GLAST-GDNF^−/− cKO mice, by crossing floxed GDNF (GDNF^f/f) mice with GLAST-CreER^T2 driver line (102). This study showed the effects of endogenous astrocytic GDNF on neuronal degeneration, brain infarction, reactive astrogliosis, oxidative stress and behavioral deficits after PT. This study provided several novel findings on the GLAST-GDNF^−/− cKO mice and PT-induced FIS mouse model: (1) GDNF is highly upregulated in the brain and RAs after PT, and the increased level was maintained within a sustained time during the post-FIS period, suggesting that GDNF is a encoded by a stress-responsive gene. In addition, the secreted GDNF could act on neurons to mediate protective effect after FIS. (2) Deletion of GDNF in astrocytes was shown to exacerbate brain and hippocampal damage and neuronal death. Brain infarction and neuronal death at PIR were significantly increased in GLAST-GDNF^−/− cKO mice. These results are consistent with previous study on a MCAO rat model showing that application of recombinant GDNF could significantly reduce both infarct size and brain edema (108). (3) RAs-derived GDNF could also affect reactive astrogliosis after PT. Using confocal imaging on immunostained Brdu and Ki67, the GLAST-GDNF^−/− cKO mice exhibited a significant reduction in Brdu+ and Ki67+ proliferating cells in the PIR after PT as compared with the WT mice. Here, the GLAST-GDNF^−/− cKO mice also showed less GFAP+ RAs and lower GFAP immunofluorescent signals than the WT mice, especially at 2 and 4 days after PT. Moreover, the GLAST-GDNF^−/− cKO mice showed less Brdu+ cells and GFAP+Brdu+ double positive cells in proliferating RAs, lower ratio of GFAP+ proliferating RAs among total proliferating cells (i.e., GFAP+Brdu+/Brdu+ ratio), and lower ratios of GFAP+ proliferating RAs (i.e., GFAP+Brdu+/GFAP+ ratio) as compared with the control mice after PT. However, cell proliferation became largely stabilized at 14 days after PT, and after this time, no major differences were observed between the control and cKO mice. These results indicate that the RAs-derived GDNF can affect the dynamics of reactive astrogliosis through a cell autonomous manner (102). These results also corroborated with the data showing reduced adult neurogenesis in the GLAST-GDNF^−/− cKO mice under normal conditions, suggesting that reduced reactive astrogliosis can lead to worsening motor function recovery in the cKO mice. (4) GLAST-GDNF^−/− cKO mice showed reduced adult neurogenesis in the hippocampus in normal brain. In addition to having proliferative capacity, RAs also exhibit many other stem cell-like properties (137-144). For example, they express neural stem cell markers such as nestin, Sox2 and DCX (141, 143, 145, 146). Given that GLAST is also expressed in neural stem cells, Zhang et al. examined the effects of GDNF deletion in astrocytes on adult neurogenesis in the dentate gyrus (DG) (102). Normally, neurogenesis in the adult brain exists in two neurogenic regions in the brain, namely, the subgranular zone (SGZ) of DG in the hippocampus, and the subventricular zone (SVZ) along the walls of the lateral ventricle (147, 148). Unique microenvironments (i.e., ‘niches’) in these two regions include soluble factors, membrane-bound molecules and extracellular matrix, permitting stem cells to self-renewal and progenitor differentiation into neurons in adult stage. Neurogenesis and survival of newly generated neurons can be regulated by various factors. GLAST is an astrocyte-specific glutamate transporter and is expressed in type I neural stem cell. GLAST-lineage radial glia-like cells (RGCs) can contribute to adult hippocampal neurogenesis (136). In the GLAST-GDNF^−/− cKO mice, GDNF was reduced in hippocampus under normal conditions albeit without statistical significance(102). Study by Zhang et al. also found that the cKO mice exhibit reduced adult neurogenesis in the DG under normal conditions, suggesting that GDNF in GLAST-lineage RGCs is an important growth factor to facilitate adult neurogenesis. The results of neurogenesis and GDNF levels are not contradictive, since in normal adult brain, GDNF is mostly expressed in neurons and only very low in astrocytes (93-95). On the other hand, growing evidence suggests that RAs play an important role in neurogenesis in the SVZ and SGZ in DG (149, 150). Ischemic stroke can cause substantial reactive astrogliosis in SVZ, and there is an association between endogenous neurogenesis and improved stroke outcomes (151). Therefore, RAs can play a role in stroke recovery through modulating neurogenesis via GDNF. Consistently, intracerebral infusion of GDNF was shown to promote striatal neurogenesis after stroke (107). (5) Deletion of GDNF in astrocyte could decrease G6PD and increase ROS production. Using double immunostaining of G6PD and GFAP, study by Zhang et al. showed that G6PD was largely upregulated in RAs after PT, but deletion of GDNF in astrocytes suppressed the upregulation of G6PD in RAs (102). In vivo DHE labeling confirmed that the GDNF deficient mice have much higher levels of ROS in the PIR than WT mice, further suggesting that through activating an antioxidant mechanism via PPP, RAs-derived GDNF can exert a neuronal and brain protective effect after FIS (102). These results are consistent with the in vitro and in vivo studies on Parkinson’s disease models (125, 152), showing that through analysis of oxidative stress markers such as protein carbonyls and 4-hydroxynonenal, astrocytes overexpressing GDNF or recombinant GDNF can reduce oxidative stress in 6-hydroxydopamine (6-OHDA)-induced mouse Parkinson’s disease model (152). Both markers were significantly reduced in GDNF treated animals. In addition, GDNF was shown to protect against loss of striatal and nigral tissue levels of dopamine. Taken together, these results agree with the notion that the protective effects of GDNF against 6-OHDA involves a reduction in oxidative stress. (6) GLAST-GDNF^−/− cKO mice exhibited higher motor function deficits than the control mice after PT, suggesting that RA-derived GDNF may play an important role in intrinsic brain repair and recovery in the chronic phase after FIS. Taken together, these results further demonstrate that astrocytes- and/or RAs-derived GDNF is beneficial to protect against neuronal death and brain damage after FIS, stimulate reactive astrogliosis, promote anti-oxidative stress mechanism, and improve long-term stroke outcomes. These findings firmly indicate the role of endogenous GDNF in astrocytes/RAs as an essential factor in brain repair processes.

RAs-derived GDNF may promote neuronal survival through inhibition of microglia activation. Microglial activation in stroke has been implicated as a major source of inflammation that contribute to neuronal dysfunction and degeneration (19). Factors secreted by astrocytes have shown to reduce the phagocytic activity and the production of ROS by activated microglia. Using Zymosan A-stimulated midbrain microglia cultures, Sandra et al. showed that astrocytes conditioned media was capable of modulating microglial activation, including increase in phagocytic activity and production of ROS (89). Both parameters were diminished in microglia culture incubated with astrocytes conditioned media, but not in conditioned media silenced with GDNF. These results indicate that astrocyte derived GDNF can play a major role in the control of microglial activation in stroke, and that GDNF can protect against neurodegeneration through inhibition of neuroinflammation.

Since GDNF is a potent neurotropic factor capable of promoting neuronal synaptic formation (45), it is also expected that RAs-derived GDNF may directly stimulate synaptogenesis and enhance function of surviving neurons in the PIR after FIS. However, information in this area is limited and future studies involving electrophysiology and in vivo two-photon imaging on dendritic spines may shed new lights.

4. Concluding Remarks

Evidence from clinical studies suggest that the brain is ‘primed’ to recovery during the subacute phase of ischemic stroke (153, 154). Therefore, understanding its mechanism will help to identify potential strategies to facilitate brain recovery and improve stroke outcomes. In this respect, GDNF released from RAs may offer multiple functions to mediate recovery of damaged neurons after FIS. Studies on glia-neuron interactions in health and disease reach an inflection point in which cell-specific in vivo approaches are beginning to validate evidence obtained from pharmacological approaches. With recent advance in genetic tools, such as availability of astrocyte specific transgenic mouse lines and astrocyte specific viral transduction, it is possible to look at exclusive roles of astrocytes/RAs in neuronal circuitry during neuronal degeneration/death. Further studies on the role and mechanism of GDNF in RAs may help to better understand the intrinsic brain repair mechanisms through the non-cell and cell autonomous effects, and in the context of neuron-glia interactions. These observations further demonstrate that targeting RAs can be a potentially important strategy in stroke restorative therapies.