Don't fence me in: Harnessing the beneficial roles of astrocytes for spinal cord repair

Abstract

Astrocytes comprise a heterogeneous cell population that plays a complex role in repair after spinal cord injury. Reactive astrocytes are major contributors to the glial scar that is a physical and chemical barrier to axonal regeneration. Yet, consistent with a supportive role in development, astrocytes secrete neurotrophic factors and protect neurons and glia spared by the injury. In development and after injury, local cues are modulators of astrocyte phenotype and function. When multipotent cells are transplanted into the injured spinal cord, they differentiate into astrocytes and other glial cells as opposed to neurons, which is commonly viewed as a challenge to be overcome in developing stem cell technology. However, several examples show that astrocytes provide support and guidance for axonal growth and aid in improving functional recovery after spinal cord injury. Notably, transplantation of astrocytes of a developmentally immature phenotype promotes tissue sparing and axonal regeneration. Furthermore, interventions that enhance endogenous astrocyte migration or reinvasion of the injury site result in greater axonal growth. These studies demonstrate that astrocytes are dynamic, diverse cells that have the capacity to promote axon growth after injury. The ability of astrocytes to be supportive of recovery should be exploited in devising regenerative strategies.

1. Introduction

Astrocytes are the predominant glial cell type in the adult mammalian central nervous system (CNS), comprising the majority of all glial cells, which outnumber neurons approximately ten to one (reviewed in Magistretti & Ransom, 2002; O'Kusky & Colonnier, 1982). Named originally for their star shape revealed with silver stains (reviewed in Garcia-Marin et al., 2007), the principle characteristics that have defined astrocytes are their stellate morphology, contribution to a basal lamina around blood vessels and meninges (Gotow & Hashimoto, 1988; Struckhoff, 1995), and the expression of intermediate filaments composed of glial fibrillary acidic protein, or GFAP (Eng et al., 1971). The modern definition of astrocytes continues to expand with the discovery of molecular markers that can be used to follow cell lineages in vitro and in vivo. According to the current wisdom, astrocytes comprise a heterogeneous population of cells with diverse morphological features and patterns of protein expression, varying widely with their developmental state and distribution in the CNS (reviewed in Bachoo et al., 2004; Reichenbach & Wolburg, 2005). This phenotypic diversity is also thought to reflect functional heterogeneity of these cells. Thus, considerable effort has been made to more clearly define subpopulations of astrocytes and their functions based upon the patterns of expression of cellular markers during development, normal mature nervous system functioning, and in pathological/neurological conditions (Table 1). Understanding the origin and differentiation patterns of astrocyte lineage cells is important for optimizing cell selection and pre-differentiation paradigms for transplantation strategies. In addition, exploiting the potential of astrocyte diversity may enable the development of alternative therapies that enhance endogenous cellular repair.

Table 1.

Reported marker expression patterns on different astrocyte subtypes. Symbols indicate relative expression levels.

Astrocytic Marker Expression
Antibody	Antigen Binding Site	Radial Glia	Immature Astrocytes	Fibrous Astrocytes	Protoplasmic Astrocytes	Reactive Astrocytes
Nestin	Intermediate filament protein	+ 1,2	+ 1,17	-	-	++ 28
3CB2	Intermediate filament-associated protein	++ 1	+ 1	+ 1	-	?
RC2	Intermediate filament protein	++ 6	+ 14	+ 14	-	?
Vimentin	Intermediate filament protein	++ 1,3	++ 1,11,12	+ 16	-	+++ 29
Brain lipid binding protein (BLBP)	Notch signaling protein	++ 1	+ 1	+ 1	+ 1	?
GLAST	Glutamate transporter	+ 1,4	+ 1,4	+ 13	+ 13	++ 26,27
GLT-1	Glutamate transporter	+ 5	+ 5,13	+ 13	+ 13	? *27
A2B5	Ganglioside epitope	+ 9	+ 10	+ 22	-	?
Glial fibrillary acidic protein (GFAP)	Intermediate filament protein	- **1	+ 1,15,16	++ 1,20,21	++ 1,20,21	+++ 23
S100β	Ca2+ related protein	+ ***7,8	+30	+ 8,18,19	+ 8,18,19	++ 24,25

- No Expression

+ Low Relative Expression

++ Medium Relative Expression

+++ High Relative Expression

? Unknown

^*Data Contradictory

^**Very Low Levels

^***Cerebellar Cells Only

⁽¹⁾References: Barry & McDermott, 2005;

⁽²⁾Tohyama et al., 1992;

⁽³⁾Dahl et al., 1981;

⁽⁴⁾Shibata et al., 1997;

⁽⁵⁾Furuta, et al., 1997;

⁽⁶⁾Misson et al., 1988;

⁽⁷⁾Hacem et al., 2007;

⁽⁸⁾Vives et al., 2003;

⁽⁹⁾Hirano & Goldman, 1988;

⁽¹⁰⁾Aloisi, et al., 1992;

⁽¹¹⁾Joosten & Gribnau, 1989;

⁽¹²⁾Pixley & deVellis, 1984;

⁽¹³⁾Sims & Robinson, 1999;

⁽¹⁴⁾Edwards et al., 1990;

⁽¹⁵⁾Voigt, 1989;

⁽¹⁶⁾Yang et al., 1993;

⁽¹⁷⁾Zerlin, et al., 1995;

⁽¹⁸⁾Ludwin et al., 1976;

⁽¹⁹⁾Steiner et al., 2007;

⁽²⁰⁾Bullon et al., 1984;

⁽²¹⁾Eng et al., 1971;

⁽²²⁾Scolding et al., 1999;

⁽²³⁾Jjayan et al, 1990;

⁽²⁴⁾Corvino et al., 2003;

⁽²⁵⁾do Carmo Cunha et al., 2007;

⁽²⁶⁾Beschorner et al., 2007;

⁽²⁷⁾Vera-Port ocarrero et al., 2002;

⁽²⁸⁾Clarke et al., 1994;

⁽²⁹⁾Kiyake et al, 1988;

⁽³⁰⁾Brusco et al., 1995.

2. Astrocyte development and heterogeneity

In the developing nervous system, astrocyte precursors and progenitors are identified by their morphology and the patterns of expression of a number of cellular proteins that serve as markers of differentiation (Fig. 1). Astrocytes arise from precursors called radial glia that are generated in the neuroepithelium (reviewed in Malatesta et al., 2008; Morest & Silver, 2003). Radial glia are defined by their morphology as elongated, bipolar cells that extend processes through the tissue, from the ventricular zone to the pia. Radial glia give rise to neurons (Malatesta et al., 2000) and astrocytes, and they express a specific pattern of cellular markers as they proceed in development (Table 1). During neurogenesis, all neural progenitor cells, including the radial glia, express the intermediate filament proteins nestin (Barry & McDermott, 2005; Tohyama et al., 1992) and vimentin (Barry & McDermott, 2005; Dahl et al., 1981). After neurogenesis is completed, radial glial cells that will go on to differentiate into astrocytes continue to express nestin and vimentin, but begin to express brain lipid binding protein (BLBP) (Barry & McDermott, 2005) in response to activation by Notch signaling (Anthony et al., 2005). In contrast, the elongated glial precursors that go on to form neurons do not express BLBP. Expression of BLBP is thus commonly used as a specific marker for the identification of radial glial astrocyte progenitors (Barry & McDermott, 2005). In addition, they also begin to express the glutamate transporter, GLAST (EAAT1) (Barry & McDermott, 2005; Shibata et al., 1997). The other glutamate transporter expressed on astrocytes, GLT-1 (EAAT2), is observed in the spinal cord shortly after the appearance of GLAST (Furuta et al., 1997). This is also when RC2, a common marker for radial glial cells that identifies a 295 kDa intermediate filament-associated protein (Chanas-Sacre et al., 2000), is expressed in the developing spinal cord (Misson et al., 1988). Mouse radial glial cells in the cerebellum express S100β (Hachem et al., 2007; Vives et al., 2003), but this expression has not yet been shown in other radial glia populations.

Fig. 1.

Astrocyte differentiation in development and after spinal cord injury. Radial glial cells (BLBP+) will give rise to immature astrocytes. Later in development, these transitional cells will differentiate into either protoplasmic (grey matter) or fibrous (white matter) astrocytes. After spinal cord injury, ependymal cells can differentiate into GFAP+ astrocytes. Quiescent astrocytes will exhibit hypertrophied processes and increase expression of cytoskeletal proteins, becoming reactive astrocytes.

As development proceeds, BLBP+ radial glia maintain a bipolar morphology and typically retain contact with the pial surface, but lose their basal attachment to the ventricular zone. At this point they undergo translocation of the nucleus toward the periphery and they initiate the expression of a filament associated protein recognized by 3CB2 (Barry & McDermott, 2005), the ganglioside epitope A2B5 (Hirano & Goldman, 1988), and very low levels of GFAP (Barry & McDermott, 2005). Prior to their differentiation into astrocytes, the radial glial cells continue to provide structural and metabolic support for the migration and differentiation of developing neurons as they proceed to the target areas within the neuropil (Rakic, 1978; reviewed in Vaccarino et al., 2007). In addition, consistent with a supportive role for developing neurons, the radial glia continue to express the primary glutamate transporters, GLAST (Shibata et al., 1997) and GLT-1 (Furuta, Rothstein, & Martin, 1997), and neurotrophic factors, such as glial-derived neurotrophic factor (GDNF), that are essential for survival of immature neurons (Koo & Choi, 2001).

After reaching their destination in gray or white matter, radial glial cells are further influenced by local secreted factors and cell adhesion molecules. Due to the difficulty of studying these mechanisms in vivo, the specific dynamics are still poorly understood, but effects have been studied extensively on astrocyte-lineage cells in vitro. A few relevant molecules that have been found to influence astrocyte differentiation include, but are not limited to: LIF (leukemia inhibitory factor) and BMP (bone morphogenic protein) (Bartlett et al., 1998; Li & Grumet, 2007), reelin (Hartfuss et al., 2003), aggrecan (Domowicz et al., 2008), EGF (epidermal growth factor), insulin, IGF-1 (insulin-like growth factor-1), and FGF-2 (fibroblast growth factor) (Bramanti et al., 2007). In response to these signals, transcription pathways are induced that favor the differentiation of BLBP+ radial glia into astrocytes. The radial glia retain their bipolar morphology and expression of A2B5 (Aloisi et al., 1992), nestin (Barry & McDermott, 2005; Zerlin et al.,), 3CB2 (Barry & McDermott, 2005), vimentin (Barry & McDermott, 2005; Joosten & Gribnau, 1989; Pixley & deVellis, 1984), GLAST (Barry & McDermott, 2005; Shibata et al., 1997), BLBP (Barry & McDermott, 2005), GLT-1 (Furuta, Rothstein, & Martin, 1997; Sims & Robinson, 1999) and RC2 (Edwards, Yamamoto, & Caviness, Jr., 1990). Then, they form a small number of additional processes and upregulate their expression of the prominent intermediate filament of astrocytes, GFAP (Barry & McDermott, 2005; Voigt, 1989; Yang et al., 1993). They also begin to express S100β (Brusco et al., 1995), which is often used as a marker of astrocyte-restricted cells in vitro (Raponi et al., 2007), although no published evidence shows expression of S100β in GFAP+/vimentin+ astrocytes in vivo in the immature spinal cord. This morphological change marks their progression to the immature astrocyte phenotype, one often described as being the transition between radial glia and mature astrocytes. Similar to radial glial cells, immature astrocytes are supportive for neuronal development and help to guide growing axons in the developing spinal cord (Cole & Lee, 1997; Joosten et al., 1989; McDermott, Barry, & McMahon, 2005).

In postnatal stages of development, the immature astrocytes lose their bipolar morphology, downregulate expression of vimentin (Joosten et al., 1989) and nestin (Barry & McDermott, 2005), and further increase expression of GFAP (Barry & McDermott, 2005; Bullon et al., 1984; Eng, Vanderhaeghen, Bignami, & Gerstl, 1971). They retain expression of GLAST and GLT-1 (Sims et al., 1999) throughout their processes, BLBP in the soma and distal processes (Barry & McDermott, 2005), and S100β (Ludwin, Kosek, & Eng, 1976; Steiner et al., 2007; Vives et al., 2003). Mature astrocytes are subclassified by their morphology and location, as either fibrous (white matter) or protoplasmic (grey matter) astrocytes. Fibrous astrocytes have fewer, thinner, and longer processes than protoplasmic astrocytes and continue to express A2B5 (Scolding, Rayner, & Compston, 1999), 3CB2 (Barry & McDermott, 2005), RC2 (Edwards et al., 1990), and vimentin (Yang et al., 1993), whereas protoplasmic astrocytes downregulate these markers. To date, little is known about how these differences reflect functional diversity, but the observations suggest that the local environment strongly dictates even basic phenotypic characteristics of this class of cells.

3. Role of astrocytes in the mature CNS

Mature astrocytes are vital to the maintenance and operation of the adult nervous system. Astrocytes are located in all areas of the nervous system and have different morphologies and functions depending on their location. Examples of astrocytes with specialized morphology include Müller cells, found in the retina; Bergmann glia, located in the cerebellum; tanycytes, found in the periventricular organs; and the protoplasmic and fibrous astrocytes in the white and grey matter of the brain and spinal cord, respectively (reviewed in Reichenbach & Wolburg, 2005).

Despite their differences, astrocytes in all regions have similar roles as nervous system support cells. They contribute to synaptic transmission by modulating the levels of the excitatory neurotransmitter glutamate in brain and spinal cord synapses (reviewed in Anderson & Swanson, 2000; Norenberg, 1979; Schousboe et al., 2004; reviewed in Tanaka, 2007), spatially buffering excess potassium in the extracellular space after action potentials have occurred (reviewed in Walz, 2000; Walz, Wuttke, & Hertz, 1984), and regulating neuronal calcium levels (Parpura et al., 1994). In addition, astrocytic endfeet form and maintain the blood-brain barrier (reviewed in Abbott et al., Hansson, 2006; reviewed in Haseloff, Blasig, Bauer, & Bauer, 2005; Vise, Liss, Yashon, & Hunt, 1975), and astrocytes play a key role in energy (ATP) production and metabolism by transporting glucose and converting carbon dioxide into water and protons (Hertz, Drejer, & Schousboe, 1988; reviewed in Jakovcevic & Harder, 2007). These protons are then released into the extracellular space and help to maintain the optimal pH of the nervous system. Finally, astrocytes contribute to volume homeostasis via the water transport channel aquaporin 4 (AQP4) (Nielsen et al., 1997). Thus, normal astrocyte structure and functions are essential for nervous system development and mature physiological activity.

The necessity of functional astrocytes is supported by the consequences of deletion of the major intermediate filament protein in astrocytes. Mice lacking a functional GFAP gene exhibit impaired myelination, hydrocephalus, and abnormal blood-brain-barrier functioning (Liedtke et al., 1996). Humans with a mutated gene encoding GFAP develop Alexander disease, a fatal neurogenerative disorder. Patients suffering from Alexander disease have multiple cognitive and motor deficits, such as impaired mental development, seizures, and difficulty with balance and speaking, in addition to a shortened lifespan (reviewed in Li, Messing, Goldman, & Brenner, 2002; Quinlan et al., 2007).

4. Role of astrocytes after CNS trauma

Astrocytes exhibit phenotypic characteristics and responses to spinal cord injury depending on location in the spinal cord and proximity to the site of injury or degeneration. In the vicinity of a traumatic injury, environmental cues associated with cell damage and neuroinflammation induce astrocytes to undergo hypertrophy, proliferate, migrate, differentiate, and form a dense network bordering the lesion site. This response contributes to the formation of the glial scar, traditionally viewed as both a physical and chemical barrier to regeneration (Liuzzi & Lasek, 1987; Reier, Stensaas, & Guth, 1983; Rudge & Silver, 1990). In addition, at sites of contact between astrocytes and peripheral cells, such as Schwann cells and fibroblasts, a glial limitans may be produced, which can effectively function as a fence guiding CNS axons around, but preventing growth of axons into the lesion site (Matthews, St Onge, Faciane, & Gelderd, 1979).

In addition to changing morphology after injury, astrocytes also alter their protein expression. They increase expression of GFAP (Vijayan, Lee, & Eng, 1990), S100β (Corvino et al., 2003; do Carmo Cunha et al., 2007), GLAST (Beschorner et al., 2007; Vera-Portocarrero et al., 2002), and GLT-1 (Vera-Portocarrero et al., 2002), and re-express many of the markers found during their development, including nestin (Clarke et al., 1994) and vimentin (Miyake, Hattori, Fukuda, Kitamura, & Fujita, 1988) (Fig. 2). In addition, they increase expression of inhibitory extracellular matrix molecules such as chondroitin sulfate proteoglycans (CSPGs) and ephrins, which can inhibit growth into the lesion as well as growth past the lesion site into the denervated CNS (Fitch & Silver, 2008; Morgenstern, Asher, & Fawcett, 2002; Zuo et al., 1998). Finally, astrocytes produce chemokines that attract macrophages from the periphery to the site of injury (Otto et al., 2002; Strack et al., 2002).

Fig. 2.

Astrocytes express different markers dependent on the region of spinal cord. (A) GFAP, vimentin, and nestin are expressed in reactive astrocytes on the lesion border. (B) Cells surrounding the central canal distal from the epicenter express vimentin and nestin, but not GFAP. (C) Spared gray matter astrocyte distal from the epicenter express only GFAP. (D) Spared white matter cells distal from the epicenter express GFAP, vimentin, and nestin. Bar = 10 μm.

Astrocytes that undergo these changes at the lesion site are considered to be “reactive”. In contrast, astrocytes that are located at areas distal to the injury respond to different environmental cues. They express increased 3CB2 (Shibuya et al., 2003), GFAP, vimentin, and nestin (Fig. 2), and are said to be “activated” (reviewed in Liberto, Albrecht, Herx, Yong, & Levison, 2004). However, these more distant astrocytes do not contribute to the formation of the glial scar and do not produce an environment that is inhibitory to axonal elongation (Davies et al., 1999). For the purposes of this review, the term “reactive astrocytes” will refer to those cells that form the glial scar around the injury site.

Reactive astrocytes can arise either from astrocytes that are already present at the time of injury, or from progenitor cells that are found either in regions surrounding the central canal (Beattie et al., 1997) or the subpial region of the spinal cord (Wu et al., 2005). A recent fate mapping study has shown that quiescent, mature astrocytes can also proliferate and give rise to astrocytes that contribute to the glial scar after a stab wound in the brain (Buffo et al., 2008). These cell populations are considered to be adult stem cells that proliferate and differentiate into glial cells, including astrocytes, after injury (Mothe & Tator, 2005). As mentioned above, several environmental cues present after injury can contribute to the phenotypic shift of quiescent astrocytes or progenitor cells to reactive astrocytes. Immediately after spinal cord injury, the proinflammatory cytokines, interleukin-1β (IL-1β) and tumor necrosis factor (TNF) are increased, which precedes increased expression of GFAP (Pineau & Lacroix, 2007). Leukemia inhibitory factor (LIF) is increased within hours after spinal cord injury (Pineau & Lacroix, 2007), which also increases astrocytic activation (Kerr & Patterson, 2004). Ciliary neurotrophic factor (CNTF), a nerve growth factor that is upregulated approximately three days after injury in mice (Zai, Yoo, & Wrathall, 2005), has been shown to increase astrocyte expression of GFAP and astrocyte proliferation (Ishii et al., 2006; Levison, Ducceschi, Young, & Wood, 1996). Fibroblast growth factor-2 (FGF-2), a growth factor involved in proliferation and differentiation that is increased at approximately the same time (Zai, Yoo, & Wrathall, 2005), appears to increase the reactivity of astrocytes, but not affect proliferative ability (Santos-Silva et al., 2007). In addition, there has been some evidence that activation of the epidermal growth factor receptor (EGFR) by ligands such as epidermal growth factor (EGF) and transforming growth factor-α (TGF-α) contributes to the reactivity and/or proliferation of astrocytes (Isono et al., 2003; reviewed in Liu & Neufeld, 2007; Rabchevsky et al., 1998), but this topic remains controversial as conflicting data has also emerged (Mandell, Gocan, & Vandenberg, 2001).

Although astrocytes play a key role in forming the inhibitory glial scar, the astrocyte response to injury is multifaceted and includes several actions that contribute to endogenous neuroprotection and repair. Astrocytes produce neurotrophic factors, such as brain-derived neurotrophic factor (BDNF) (Ikeda et al., 2001), CNTF (Lee et al., 1998), nerve growth factor (NGF) (Krenz & Weaver, 2000), and FGF-2 (do Carmo Cunha et al., 2007). Furthermore, astrocytes play a key role in reducing excitotoxic death after injury by upregulating their expressing of GLAST and GLT-1 as they become more reactive (Rothstein et al., 1996). In addition, astrocytes increase expression of the water transport channel AQP4 after spinal cord injury (Nesic et al., 2006). Finally, although reactive astrocytes produce CSPGs, molecules that are inhibitory to neurite growth, they also produce extracellular matrix molecules that are supportive to axon growth, such as laminin (Costa et al., 2002; Frisen et al., 1995) and fibronectin (Tom, Doller, Malouf, & Silver, 2004). These actions demonstrate the roles of astrocytes in reconstructing and protecting the nervous system after injury. Their re-expression of developmental markers is suggestive of a phenotypic shift to a morphology that is supportive to regeneration. Indeed, cocultures of neurons and vimentin+/nestin+/GFAP- astrocytes exhibit increased axonal growth, suggesting that this phenotype is one that encourages regeneration (Menet et al., 2000).

5. Modifying the astrocyte response to injury

Due to their contribution to the glial scar, astrocytes have often been targeted as cells that should be inhibited or destroyed after injury in order to reduce the scar and promote regeneration. However, studies that attempt to eliminate or manipulate astrocytes after injury reveal that their presence is needed for successful repair. This is demonstrated by several studies using GFAP-deficient mice. It has been found that animals lacking GFAP have increased hippocampal degeneration after brain injury (Otani et al., 2003), form an abnormal glial scar after brain and spinal cord lesions (Pekny et al., 1999), produce astrocytes with impaired migratory abilities in vitro (Lepekhin et al., 2001), and are more susceptible to death after cervical spinal cord injury (Nawashiro et al., 1998). In addition to studies on GFAP-deficient mice, research has been conducted using AQP4-null mice. Astrocytes from these mice have an impaired astrocyte migration and wound healing response in vitro and in vivo after a cortical stab injury (Saadoun et al., 2005). Furthermore, AQP4-null mice have increased seizure activity and brain swelling after vasogenic edema, speculated to be a result of decreased potassium reuptake by astrocytes (Verkman et al., 2006). Similar results have been exhibited in studies that interfere with astrocyte proliferation and migration to the lesion site. Ablation of proliferating astrocytes after both spinal cord and brain injury leads to impaired blood-brain-barrier repair, decreased remyelination, increased leukocyte infiltration, neuronal degeneration, and increased motor deficits (Bush et al., 1999; Faulkner et al., 2004; Myer, Gurkoff, Lee, Hovda, & Sofroniew, 2006). In addition, a conditional transgenic deletion or inhibition of STAT3, a transcription factor necessary for astrocyte migration, results in exacerbated functional deficits after injury (Okada et al., 2006).

In addition to the role of reactive astrocytes in restoring tissue integrity (Mathewson & Berry, 1985), evidence suggests that astrocytes present at the site of injury contribute to improved recovery and enhance axonal regeneration. For example, astrocyte expression of Glut-1, a glucose transporter, is needed for revascularization after injury (Whetstone et al., 2003). Furthermore, after ethidium bromide-induced demyelination in the spinal cord, endogenous oligodendrocyte precursor cells are unable to remyelinate axons in the absence of astrocytes (Talbott et al., 2005). After spinal cord injury in young rats in which the corticospinal tract (CST) shows modest regeneration, the presence of GFAP+/vimentin+ astrocytes in the lesion is necessary for axon growth (Iseda et al., 2004). Notably, transplantation of collagen tube grafts (Spilker et al., 2001), collagen gels (Joosten, Bar, & Gispen, 1995) or Schwann cells with IN-1 anti-myelin antibodies (Guest et al., 1997) result in a migration of astrocytes into the site of a spinal cord transection and support the growth of CNS derived axons. In contrast, Schwann cell grafts that exclude astrocytes are not invaded by CNS axons unless they are supplemented by exogenous neurotrophic factors (Xu et al., 1995). This suggests that with appropriate stimulation, the reactive response of astrocytes after injury could be exploited for successful endogenous repair.

6. Neural stem cell transplantation

Multipotent stem cells represent a promising source for cell replacement therapy after SCI. It is known that grafts of embryonic spinal cord or cortical tissue, containing multipotent glial and neuronal progenitors, integrate with the injured spinal cord, provide neuroprotection, alter the host glial scarring response, and can facilitate functional recovery (Bernstein & Goldberg, 1989; Bregman et al., 1993; Houle & Reier, 1988; Reier et al., 1988; Stokes & Reier, 1992). With more selective approaches, cells derived from proliferative regions of embryonic or adult brain and spinal cord can be grown as neurospheres or pluripotent cultures and will differentiate in vitro to form neurons, astrocytes and oligodendrocytes (reviewed in Cao, Benton, & Whittemore, 2002; Cao et al., 2001; Gritti et al., 1996; Vroemen, Aigner, Winkler, & Weidner, 2003; as cited in Weiss et al., 1996). These isolated cells have been introduced into the injured spinal cord in a number of different models (reviewed in Enzmann, Benton, Talbott, Cao, & Whittemore, 2006). However, as described above, elements in the microenvironment of the lesion, such as TNF-α, IL-1β, interferon-gamma (IFN-γ), and CNTF, typically promote differentiation of these cells into a glial lineage as opposed to a neuronal one (reviewed in Cao, Benton, & Whittemore, 2002; Cao et al., 2001; Ricci-Vitiani et al., 2006; Vroemen et al., 2003).

The majority of multipotent stem cells transplanted into the injured CNS differentiate into astrocytes or oligodendroglia. Contrary to traditional views that this differentiation is unfavorable for repair, several transplantation studies exhibiting astroglial differentiation have been shown to promote recovery of function after SCI (Cao et al., 2005; Hasegawa et al., 2005; Hofstetter et al., 2005; McDonald et al., 1999; Ogawa et al., 2002). A study transplanting human embryonic stem cells (20 weeks post-conception) into a transected rat spinal cord found that this treatment resulted in improved CST regeneration and improved motor function as assessed by the Basso-Beattie-Bresnahan (BBB) locomotor scale (Liang et al., 2006). Histological analysis revealed that these cells differentiated into astrocytes and oligodendrocytes, but not neurons. In order to determine the mechanism through which these cells improved recovery, conditioned media from the cells was applied to the injured spinal cord. Application of conditioned media also resulted in an improvement in CST regeneration and BBB score, though less robust than effects observed after cell transplantation. Based on these results, the authors speculated that the cells improved growth by a combination of nutritional or trophic support, and also by formation of a scaffold on which neurons can grow. Similarly, when stem cells from human spinal cord (8 weeks into gestation) were transplanted into the contused cervical spinal cords of marmoset monkeys, the majority of transplanted cells differentiated into astrocytes (Iwanami et al., 2005). Animals that received transplanted cells had improved bar grip and spontaneous locomotor activity scores. In another study, neural stem cells from E14 mouse cerebral cortex were transplanted into a midthoracic dorsal funiculotomy (Pallini et al., 2005). It was found that these cells primarily differentiated into GFAP+/vimentin+ astrocytes with a bipolar morphology, indicative of immature astrocytes. These cells appeared to support growth of host axons, suggesting that this astrocyte phenotype may be optimal to improve axonal regeneration and functional recovery.

Multiple mechanisms contribute to the improvements observed in stem cell transplantation research, including effects on reducing secondary injury, trophic factor production, and promotion of remyelination (reviewed in Enzmann et al., 2006). However, there are several challenges entailed in using multipotent stem cells, including the uncertainty of repair mechanisms, reproducibility of studies, potential tumorigenicity, and the ethical considerations for obtaining such cells. Thus, recent work has also focused on generating cells further along the differentiation pathway that are isolated for or driven toward a growth-permissive astrocytic lineage. This work indicates that the beneficial effects of astrocyte transplantation can be harnessed for CNS trauma.

7. Transplantation of astrocytic lineage cells

Recognizing the potential for astrocytes to provide support for cells and processes in the injured spinal cord, a few studies have specifically focused on grafting neonatally-derived cortical astrocytes into partial transection lesions of the adult spinal cord (Joosten, Veldhuis, & Hamers, 2004; Wang et al., 1995). In the first study, astrocytes were implanted either as a cell suspension or attached to a gelfoam pledget. Under these conditions, the astrocytes survived and integrated with the cells of the spinal cord. The lesions containing astrocyte preparations showed rapid constriction of the tissue scar and greater GFAP immunoreactivity. Consistent with the growth supportive nature of the astrocytes, more axons were present within the lesion site. However, the origin of these axons and the effects of the grafts on the surrounding glial environment were not fully assessed. In the latter study, neonatal astrocytes were implanted in a collagen gel that was placed in a dorsal column lesion. The authors reported no migration of transplanted cells into the host spinal cord, but found increased neurofilament staining and were able to demonstrate short distance growth of some CST axons into the rostral end of the transplant. Similarly, other studies have favored transplantation of astrocytes of a more immature phenotype in brain injuries, resulting in improved recovery (Smith, Miller, & Silver, 1986; Smith & Silver, 1988).

One approach to isolating permissive glial cells for transplantation takes advantage of the glial lineage restriction marker A2B5 to isolate embryonically derived glial restricted progenitors (GRPs) from developing rat spinal cord (Dietrich, Noble, & Mayer-Proschel, 2002; Mujtaba et al., 1999; Olby & Blakemore, 1996; Rao, Noble, & Mayer-Proschel, 1998). These cells are able to generate oligodendrocytes and astrocytes and exhibit self-renewal in vitro. Transplantation of isolated GRP cells into a contusion injury in the midthoracic rat spinal cord resulted in differentiation of those cells into astrocytes and oligodendrocytes (Hill et al., 2004). This treatment reduced chondroitin sulfate proteoglycan expression, altered the morphology of glial processes, reduced glial scar formation surrounding the lesion, and enhanced sprouting of injured CST axons. Further development of these cells has been pursued, including their immortalization by transfection with a retrovirus carrying the oncogenes c-myc (Wu et al., 2002) or T antigen (Trotter, et al., 1993). However, these immortalized cell lines favor differentiation into a mature, reactive astrocyte phenotype that could be detrimental to recovery and repair. Due to this caveat, many researchers have pursued alternative or additional strategies to select immature astrocytes for transplantation. Controlled differentiation of astrocytes is one such strategy.

The sequence of events that lead to developmental differentiation of astrocytes, described above, is highly dependent upon the temporal and spatial expression of active signaling molecules. These events have been exploited to drive progenitor cells along selected lineage pathways. One signal that is important in the differentiation pathway is the bone morphogenic protein (BMP) family of growth factors. BMP-4, together with retinoic acid, EGF and FGF-2, mediates the transition of embryonic stem cells along an ectodermal and mesodermal direction (Schuldiner et al., 2000). When embryonic neural progenitor cells are exposed to BMPs, their developmental fate proceeds along an astrocytic lineage (Grinspan et al., 2000; Gross et al., 1996; Nakashima et al., 1999), while inhibitors of BMPs, including noggin, shift progenitors to a neural lineage. Pursuing this approach to promote astrocyte differentiation, Davies et al. (2006) pre-differentiated A2B5+ GRP cells from embryonic rats with BMP-4 to generate GFAP+/A2B5- astrocytes and transplanted these cells into dorsal column-injured rat spinal cords. These BMP-4 stimulated astrocytes increased the growth of ascending dorsal column axons rostral to the lesion, aligned and reduced unhealthy phenotypes in GFAP+ host astrocytes, suppressed expression of CSPGs, increased cell number in the red nucleus, and also improved gridwalk performance. Conversely, transplantation of GFAP+/A2B5+ cells prepared by exposure to CNTF resulted in no improvement in functional recovery and an increased occurrence of allodynia (Davies et al., 2007). Similarly, although transplantation of neural stem cells that differentiated into astrocytes increased growth of axons in a similar spinal cord lesion model, this axon growth included aberrant sprouting and resulted in allodynia after spinal cord injury (Hofstetter et al., 2005). Together, these studies reiterate that astrocytes represent a heterogenous population that can have both beneficial and detrimental effects after injury, depending on how they are differentiated and what markers are expressed. In addition, these studies underscore the necessity for evaluating sensory responses following manipulations that result in increased axonal growth after SCI. In light of evidence that co-stimulatory factors including cytokines, mitogens, and other signaling molecules are regulated at different times after injury (Pineau & Lacroix, 2007), it is reasonable to speculate that astrocyte activation might yield different functional outcomes depending on the timecourse of stimulation and the factors present in the local environment (Ricci-Vitiani et al., 2006). If the underlying mechanisms can be identified and endogenous astrocytes can be selectively induced to exhibit a supportive phenotype, these manipulations have great potential to be used in treatment for spinal cord injury.

An alternative approach to derive supportive astrocytes for transplantation has been to select astrocyte precursors using other markers of the differentiation pathway. For example, Hasegawa (2005) found that transplantation of immortalized nestin+/BLBP+/GFAP- radial glial cells from E13.5 rats into the contused adult spinal cord resulted in improved functional recovery. Similar to their role in development, these radial glial cells formed bridges that functioned as scaffolds on which axons can grow. In addition, transplantation of these cells reduced inhibitory CSPGs and macrophage invasion into the lesion site. Another approach has been to use other CNS-derived cells with astrocyte lineage potential. These include tanycytes derived from the hypothalamus (Prieto, Chauvet, & Alonso, 2000) and ependymal cells from the choroid plexus (Kitada, Chakrabortty, Matsumoto, Taketomi, & Ide, 2001). In the former case, tanycyte grafts represent a promising cell source, as they generate a supportive environment with numerous neurofilament profiles in the lesion. In contrast, the choroid plexus derived cells differentiated into GFAP+/vimentin- cells and did not support axonal growth. Taken together, the results of three decades of neural transplantation studies indicate that astrocytes can be induced to provide a beneficial environment for regeneration, yet we still know very little about how to best utilize these cells for repair.

While transplantation offers hope for restoring a healthy cellular environment following SCI, there are general limitations to its therapeutic application, including establishing clear guidelines for production, administration, and safety of each cell type applied to this complex condition. An attractive alternative is to identify the signals that contribute to the endogenous repair processes inherent in astrocyte function and optimize them for promoting functional recovery after injury. The proliferative and supportive potential of astroglial precursors represents a promising target for such an approach.

8. Endogenous stem cells

In non-mammalian species that demonstrate successful spinal cord regeneration, a subset of ependymal cells proliferate and differentiate in concert with successful growth of regenerating axons (Beattie, Bresnahan, & Lopate, 1990; reviewed in Chernoff, 1996; reviewed in Chernoff, Stocum, Nye, & Cameron, 2003; Michel & Reier, 1979; Nordlander & Singer, 1978). The responsive ependymal cells in these species exhibit changes in morphology and cytoskeletal protein expression as they migrate into the lesion site. Typically, GFAP-positive cells are lost from the lesion and vimentin appears in the migrating ependymal cells until axonal regeneration is complete, after which GFAP expression returns (O'Hara, Egar, & Chernoff, 1992). Interestingly, it appears from close examination of these studies that axonal regrowth in these species precedes the front of glial cell migration at the lesion site, suggesting that the ependymal cells do not provide a preformed scaffold, but instead follow and support the growing neurites (Dervan & Roberts, 2003; Michel et al., 1979; Stensaas, 1983).

In the mammalian spinal cord, traumatic injury also induces a profound proliferative response of cells surrounding the central canal (Beattie et al., 1997; Guth et al., 1985; Mothe & Tator, 2005; Shibuya et al., 2002; Takahashi, Arai, Kurosawa, Sueyoshi, & Shirai, 2003). Ependymal and subependymal cells lining the central canal of the spinal cord are thus a possible source of endogenous multipotent cells that can proliferate and differentiate after injury. In vitro studies have shown that cells from the adult spinal cord ependymal region can differentiate into oligodendrocytes and radial glial cells (Kulbatski et al., 2007). In vivo studies have demonstrated that these cells proliferate, migrate, and differentiate into GFAP+ cells after a minimal midthoracic spinal cord injury (Mothe & Tator, 2005). It is thought that these proliferative cells respond to increased levels of growth factors, including FGF-2 and EGF, that are found at the site of injury. Similarly, migration and proliferation of injury-reactive ependymal cells in the regenerating axolotl (salamander) spinal cord was shown to be dependent on EGF in vitro (O'Hara & Chernoff, 1994).

9. Endogenous cell proliferation and repair

After SCI, normally quiescent cells surrounding the central canal and the subpial surface as well as in gray and white matter regions of the spinal cord undergo increased cell division, as detected by pulse labeling with systemic administration of bromodeoxyuridine (BrdU) (Horky, Galimi, Gage, & Horner, 2006; Lytle & Wrathall, 2007; McTigue, Wei, & Stokes, 2001; Mothe et al., 2005; Xu, Kitada, Yamaguchi, Dezawa, & Ide, 2006; Zai & Wrathall, 2005). Without intervention, many of these new cells differentiate along a glial lineage to form NG2+ oligodendrocyte precursors, oligodendrocytes, and astrocytes. After some types of injury, including cryogenic lesions and severe contusion or crush, populations of endogenous cells can form cordons or tissue bridges that support limited axon growth within the lesion site in a manner resembling the ependymal cell migration seen in lower vertebrates (Beattie et al., 1997; Brook et al., 1998; West & Collins, 1989). However, in most SCI models, the new cells join with reactive glia throughout the adjacent spinal cord to form the glial scar at the edge of the lesion site. Many of these cells and processes are nestin and GFAP positive (Frisen et al., 1995), but the contribution of reactive cells and cells derived from progenitors within the ependymal zone or subpial regions has been difficult to establish. The dual nature of the progenitor and reactive astrocyte response to injury has led to two camps of thought regarding the most effective approach to exploit the endogenous glial response and enhance repair.

One view is based on understanding that the endogenous formation of the glial scar is detrimental to recovery through the formation of a fence or barrier to regeneration. Although the astrocyte response is essential for restoration of the CNS environment (Faulkner et al., 2004), there is increasing evidence that reducing cell proliferation in the acute phase after injury with x-irradiation may be beneficial (Kalderon & Fuks, 1996; Ridet et al., 2000; Zeman et al., 2001; Zhang, Geddes, Owens, & Holmberg, 2005). More recently, cell cycle inhibitors such as olomoucine have been applied to reduce scar formation in a similar manner (Tian et al., 2007). However, both of these approaches are directed at all proliferating cells, including inflammatory cells, and they appear to have a primary effect on neuroprotection through their inhibition of proliferation of other cell types, not necessarily astrocytes.

In a seemingly opposite approach, other researchers have suggested augmenting the proliferative response of endogenous astrocyte progenitors by administration of growth factors or signals that will increase progenitor proliferation and migration. Intrathecal infusion for 14 days after compression injury in rats with the glial mitogens and differentiation factors FGF-2 and EGF resulted in increased glial proliferation and improved functional recovery (Kojima & Tator, 2002). However, the anatomical substrate for recovery in these studies is not clear, and further examination of this combination is needed. Other factors may be more appropriate for the ultimate goal of enhancing differentiation of cells toward a supportive phenotype as has been done with embryonic precursors for transplantation (Davies et al., 2006). In our laboratory, we have been interested in this approach based on observations of variation in the cellular response to injury in different mouse strains. Following contusion injury in mice, the lesion is characterized by fibrosis and increased cellularity (Jakeman et al., 2000; Kigerl, McGaughy, & Popovich, 2006; Kuhn & Wrathall, 1998; Ma, Basso, Walters, Stokes, & Jakeman, 2001; Sroga, Jones, Kigerl, McGaughy, & Popovich, 2003). There is considerable variation in the cellular composition within the lesion site across different strains of mice. Of the strains we have examined, including C57BL/6, C57BL/10, Balb/c, MRL/MpJ, 129S4/SvE and 129X1/SvJ, the last strain is the only one to date in which astrocytes reinvade the lesion site in a dramatic fashion (Fig. 3A–D). Between 1 and 2 weeks following a contusion injury, GFAP+ astrocytes and axons extend into the lesion site (Ma et al., 2004). The events are accompanied by a reduced macrophage response, but subsequent strain analyses indicate that reduced inflammation alone does not induce astrocyte reinvasion or axon growth (Kigerl et al., 2006). Axon growth into the lesion is also observed in strains with high Schwann cell invasion, but only in the 129X1/SvJ mice are centrally derived axons, identified by expression of 5-HT, seen within the lesion core (Fig. 3E,F). When BrdU is administered in the first week after injury, the lesion site is occupied by numerous BrdU+ astrocytes, supporting the interpretation that these represent a population of supportive astrocytes born after injury. Many of the serotonergic axons are associated with astrocyte processes, indicating that the proliferation of astrocytes and reinvasion or migration into the site of injury is a very promising approach to promoting repair and regeneration of centrally derived axons after spinal cord injury.

Fig. 3.

The lesion site from two mouse strains is different at 6 weeks after a contusion injury with no further intervention. A,C,E are from C57BL/6J mice. The lesion center is devoid of GFAP+ astrocyte processes and 5-HT immunoreactive fibers. B,D,F are from 127X1/SvJ mice. Astrocytes occupy the center of the lesion, where they are accompanied by numerous 5-HT immunoreactive fibers. Boxes in A, and B are regions that are enlarged in C,E and D,F, respectively. Scale bars in A and B = 100 μm; C-E = 20 μm.

10. Conclusions

Astrocytes are heterogeneous, plastic cells that are integral to healthy development and mature functioning of the central nervous system. After trauma, they respond to local environmental cues by proliferating, migrating to the site of the insult, and forming the glial scar, commonly viewed as a fence or barrier to regeneration. Because of this response, astrocytes have historically had a stigma that has led many researchers to target them as cells that should be depleted after injury in order to attenuate their presence at the lesion site. However, ablation of astrocytes has proven to be detrimental to recovery due to the numerous endogenous repair mechanisms that astrocytes execute when stimulated. Furthermore, transplantation of both multipotent stem cells that differentiate into astrocytes and lineage-restricted astroglial cells into the injured spinal cord has proven to be a successful tactic for nervous system repair. Manipulation of endogenous stem cells, many of which differentiate into astrocytes, in the adult spinal cord is proving to be a promising method of treatment. Thus, it is not simply the presence of astrocytes, but their phenotype and location, that determines permissivity for axonal growth, as illustrated in Fig. 4. These studies demonstrate that by harnessing the beneficial aspects of astrocytes, we can further advance in our pursuit of a treatment for spinal cord injury.

Fig. 4.

Astrocyte phenotype and location defines permissivity of the lesion environment. The left side of the figure shows the normal astrocyte response to injury, with hypertrophied astrocytes forming a glial scar around the lesion with increased production of CSPGs. Axons attempting to grow into the lesion stop at the scar and form endbulbs. The right side of the figure shows a manipulated astrocyte response in which astrocytes with an immature phenotype penetrate into the lesion and guide growing axons. One possible source of these cells is the ependymal zone surrounding the central canal, illustrated on the far left.