Thrombospondins as key regulators of synaptogenesis in the central nervous system

Abstract

Thrombospondins (TSPs) are a family of large, oligomeric multidomain glycoproteins that participate in a variety of biological functions as part of the extracellular matrix (ECM). Through their associations with a number of binding partners, TSPs mediate complex cell-cell and cell-matrix interactions in such diverse processes as angiogenesis, inflammation, osteogenesis, cell proliferation, and apoptosis. It was recently shown in the developing central nervous system (CNS) that TSPs promote the formation of new synapses, which are the unique cell-cell adhesions between neurons in the brain. This increase in synaptogenesis is mediated by the interaction between astrocyte-secreted TSPs and their neuronal receptor, calcium channel subunit α2δ-1. The cellular and molecular mechanisms that underlie induction of synaptogenesis via this interaction are yet to be fully elucidated. This review will focus on what is known about TSP and synapse formation during development, possible roles for TSP following brain injury, and what the previously established actions of TSP in other biological tissues may tell us about the mechanisms underlying TSP’s functions in CNS synaptogenesis.

1. Introduction

Thrombospondins (TSPs) are secreted multidomain glycoproteins found throughout the body of vertebrates and lower metazoa (Adams, 2001; Bentley and Adams, 2010; Mosher and Adams, 2012). The TSP family consists of 2 subgroups organized by oligomerization state and domain structure: subgroups A and B. Subgroup A includes the trimeric TSP-1 and TSP-2, while the pentameric TSP-3, TSP-4 and TSP-5 comprise subgroup B (Lawler, 2002). Unlike other extracellular matrix (ECM) proteins such as collagen and laminin that play structural roles in the ECM, TSPs are primarily involved in regulating cell-cell and cell-matrix interactions (Bornstein, 2000). To do so TSPs act through a number of extracellular matrix proteins and cell surface receptors (Table 1) and control cytoskeletal dynamics, cell migration and cell attachment.

Table 1.

Binding partners of TSP and their known functions in the CNS.

Receptor	TSP Interaction Site	CNS Functions
α2δ-1 (Eroglu et al., 2009)	Type 2 EGF-like repeats (Eroglu et al., 2009)	Synapse formation (Eroglu et al., 2009)
ApoER2 (Blake et al., 2008)	unknown	Reelin signaling; neuronal migration (Herz and Chen, 2006)
CD36 (Asch et al., 1992)	Type 1 repeats (Guo et al., 1997)	Microglial activation; brain lipid metabolism (Abumrad et al., 2005)
CD47/IAP (Gao et al., 1996)	C-terminal domain (Kanda et al., 1999)	Neurite development (Ohnishi et al., 2005)
Heparin (Lawler et al., 1995)	N-terminal domain (Lawler et al., 1995)	Cell-cell recognition and adhesion (Cole et al., 1986)
HSPG (Sun et al., 1989)	Type 1 repeats (Iruela-Arispe et al., 1999)	Cell adhesion; astrocyte migration (Faber-Elman et al., 1995)
Integrin (Lawler and Hynes, 1989)	N-terminal domain; Type 3 repeats (Bentley and Adams, 2010)	Neuronal migration; synapse architecture and function (Beumer et al., 2002)
Latent TGF-β (Murphy-Ullrich et al., 1992)	Type 1 repeats (Schultz-Cherry et al., 1994)	Activivation of TGF-β; Cytoskeletal stability; mobilization of synaptic machinery (Packard et al., 2003)
LRP1/CRT (Mikhailenko et al., 1995)	N-terminal domain (Goicoechea et al., 2000)	Endocytosis of MMPs (Emonard et al., 2004); Notch signaling (Kinoshita et al., 2003)
Neuroligin (Xu et al., 2010)	unknown	Synapse formation (Graf et al., 2004)
Notch (Meng et al., 2009)	unknown	Neural progenitor cell proliferation and differentiation; neuronal morphology (Ables et al., 2011)
VLDLR (Blake et al., 2008)	unknown	Reelin signaling; neuronal migration (Herz and Chen, 2006)

TSPs start to exert their effects as early as embryogenesis and are critical for the development of many organs in the body including the bones, muscles, heart and the brain (O'Shea et al., 1990a; Tucker et al., 1995). TSPs 1–4 have all been found in the brain (Iruela-Arispe et al., 1993; Lawler et al., 1993; O'Shea et al., 1990b). During early postnatal development they are primarily expressed by astrocytes, the predominant non-neuronal cell type in the CNS (Cahoy et al., 2008; Eroglu, 2009). Several studies in the last decade have highlighted astrocytes and the TSPs that they secrete as major controllers of formation of neuronal synaptic connections in the developing nervous system (Christopherson et al., 2005; Hughes et al., 2010; Ullian et al., 2001).

Astrocytes are the most abundant cell type in the brain. They were originally viewed as merely the “glue” that filled in the space between neurons (Volterra and Meldolesi, 2005), but recent studies have shown them to be far more active participants in the development, maintenance and plasticity of the CNS than was previously thought. In fact, it has even been suggested that it may be astrocytic complexity which underlies the vast functional competency of the human brain (Oberheim et al., 2006). Astrocytes are often found in close apposition to the pre- and postsynaptic machinery of neurons at the excitatory (glutamatergic) connections, an arrangement that has come to be known as the “tripartite synapse” (Araque et al., 1999). Through this close contact with neurons, astrocytes can modulate the efficacy of synapses through release and uptake of neuroactive substances (Eroglu et al., 2008). As a complex process-bearing cell, a single astrocyte may contact and potentially coordinate the activity of up to 100,000 synapses at once (Bushong et al., 2003;2004).

Besides their structural and functional associations at excitatory synapses in the adult brain, astrocytes also play important roles in the regulation of synapse formation and elimination in the developing CNS (Christopherson et al., 2005; Ullian et al., 2001). Excitatory synaptogenesis in the mammalian CNS occurs primarily after birth. In rodents this synaptogenic period is during the second and third postnatal weeks. A host of neuronal cell-surface molecules and secreted signals contribute to synaptic organization and maturation (Kennedy and Ehlers, 2006; Ziv and Garner, 2004) but the cellular and molecular interactions that initiate this synaptogenic period are largely unknown. This period of synaptic development also closely correlates with the proliferation and differentiation of astrocytes (Ullian et al., 2004; Ullian et al., 2001). To determine the role of developing astrocytes in synapse development, Barres and colleagues used a purified retinal ganglion cell (RGC) culture system. When these retinal neurons were cultured in the complete absence of astrocytes (Meyer-Franke et al., 1995) they formed very few synapses and had low synaptic activity (Pfrieger and Barres, 1997). Conversely when the RGCs were cultured with astrocyte feeder layers or culture media that were conditioned by astrocytes, they had 3–7 fold higher number of synapses and over 10 fold more synaptic activity. These results showed that synaptogenesis is not only controlled by intrinsic mechanisms of neurons but are stimulated by astrocyte-secreted prosynaptogenic signals. Further investigation identified one of these signals as none other than thrombospondin-1 (Christopherson et al., 2005), the ECM molecule whose synthesis and secretion by astrocytes had been discovered nearly two decades earlier (Asch et al., 1986).

2. Thrombospondins promote excitatory synapse formation in the CNS

Using the purified RGC culture system described above, Christopherson and colleagues (2005) found that pure TSP-1 and TSP-2 mimicked the ability of astrocyte conditioned media (ACM) to increase the number of excitatory (glutamatergic) synapses formed by RGCs in culture. Furthermore, immunodepletion of TSP-2 from the ACM prevented astrocyte-induced synaptogenesis. These results showed that TSP-1 and TSP-2 are necessary and sufficient signals coming from astrocytes that stimulate excitatory synaptogenesis between RGCs. Electron microscopy (EM) was used to show that the TSP-1-induced synapses are ultrastructurally normal, indicating that TSP-1 is able to trigger the formation and proper alignment of pre- and postsynaptic specializations. However, whole-cell recordings revealed that though the synapses were presynaptically active with cycling synaptic vesicles, they were postsynaptically silent owing to a lack of functional 2-amino-3-(5-methyl-3-oxo-1,2-oxazol-4-yl)propanoic acid (AMPA) receptors (Christopherson et al., 2005). These results hinted at the presence of additional signals in the ACM that can regulate different aspects of synaptogenesis (Ullian et al., 2004). Increased excitatory synaptogenesis in response to TSP-1 treatment was also seen in astrocyte-depleted hippocampal neuronal cultures (Hughes et al., 2010). However treatment of classical astrocyte-containing hippocampal cultures with TSP-1 did not lead to an increase in final synaptic density between hippocampal neurons, even though an increased rate of synaptogenesis during the early stages of synapse formation was observed (Xu et al., 2010). This difference is most likely due to the high levels of TSP-1 and TSP-2 that are secreted by astrocytes in the hippocampal culture (Xu et al., 2010), minimizing the impact of additional TSP-1 treatment. Secreted proteins from astrocytes were also shown to be involved in formation of inhibitory synapses that use gamma amino butyric acid (GABA) as the neurotransmitter (Elmariah et al., 2005a; 2005b), but TSP-1 was not crucial for this effect (Hughes et al., 2010).

Investigation of synaptic development in vivo in TSP-1 and TSP-2 double null mice confirmed the in vitro findings showing a role for TSPs in excitatory synaptogenesis, because TSP1/2-null mice had 30% fewer excitatory synapses in the cortex compared to their wild type littermates (Christopherson et al., 2005).

In agreement with a role in the initiation of excitatory synaptogenesis in vivo, the expression of TSP-1 and TSP-2 by astrocytes peaks at the start of the synaptogenic period in the mouse brain (postnatal days 5–10), but is decreased in the adult brain (Christopherson et al., 2005), and is restricted to areas of ongoing neurogenesis (Hoffman et al., 1994). TSP-4, which was first identified in Xenopus embryos (Lawler et al., 1993), is also expressed in endothelial cells and smooth muscle cells of the brain vasculature (Stenina et al., 2003), in mammalian astrocytes in the mature CNS (Cahoy et al., 2008), and has been associated with axonal outgrowth and adhesion of retinal ganglion cells (Dunkle et al., 2007). Additionally, it is known to be present at the neuromuscular junction (Arber and Caroni, 1995), suggesting a possible role for TSP-4 in synapse formation in the peripheral nervous system.

All 5 of the known TSP family members induce synaptogenesis in cultured RGCs (Eroglu et al., 2009), and their synaptogenic activity was mapped to the epidermal growth factor (EGF)-like repeats of the TSP molecule. These EGF-like domains were known to bind to the Von Willebrand Factor A (VWF-A)-like domain of certain integrins (Pluskota et al., 2005), so Eroglu and colleagues decided to probe other cell-surface proteins that contained the VWF-A domain to look for potential TSP binding partners. Their search led them to discover the novel interaction between TSPs (TSP-1, TSP-2 and TSP-4) and the voltage-gated calcium channel (VGCC) subunit, α2δ-1 (Eroglu et al., 2009). Excitatory neurons including RGCs express high levels of α2δ-1, which is localized to synapses. Overexpression of α2δ-1 both in vitro and in vivo led to increases in synaptogenesis, while α2δ-1 knockdown resulted in loss of TSP-1-induced synapse formation in vitro. Closer inspection of the TSP-1-α2δ-1 association revealed that the extracellular component of α2δ-1 was responsible for its synaptogenic effects, and that its synaptogenic properties did not involve global changes in calcium channel expression levels or function despite the known roles of α2δ-1 in calcium channel function and trafficking (Eroglu et al., 2009).

α2δ-1 is the high affinity receptor for two commonly prescribed anti-epileptic, anti-neuropathic pain medications, gabapentin (NeurontinTM) and pregabalin (Lyrica™) (Gee et al., 1996). Gabapentin and pregabalin were initially designed as hydrophobic gamma amino butyric acid (GABA) analogs to be used as anti-convulsants that would easily cross the blood brain barrier. However, further studies have shown that even though they possess anti-convulsant properties, they do not bind to canonical GABA receptors or transporters. A study using a knock-in mouse that expresses a mutant α2δ-1 which cannot bind gabapentin or pregabalin has shown that α2δ-1 is the in vivo target for these drugs, and that these drugs mediate their therapeutic action through binding to α2δ-1 (Field et al., 2006). The cellular mechanisms underlying the mode of action of these drugs remain a mystery, because these drugs do not significantly affect calcium channel kinetics or function. Eroglu and colleagues showed that gabapentin blocks TSP-α2δ-1 binding and astrocyte/TSP-induced synapse formation in vitro (Eroglu et al., 2009). The drug similarly blocked synapse formation in vivo. In this experiment, neonatal mice were injected with either gabapentin or saline for the first postnatal week, which coincides with the initiation of synapse formation and the peak of TSP1/2 expression in the brain. There was a significant decrease in the density of excitatory synapses in the cerebral cortex of the gabapentin-injected mice in comparison to saline-injected control mice. These results showing that gabapentin can block new synapse formation in vivo and in vitro indicated that gabapentin’s ability to block TSP/α2δ-1 interaction and TSP-induced synapse formation might underlie its therapeutic function.

2.1 Potential mechanisms of thrombospondin function at the synapse

It has been proposed that the binding between the EGF-like domains of TSP and the VWF-A-like domain of α2δ-1 may act as an “on switch” for a synaptogenic signaling pathway. This may be achieved by the activation of a “synaptogenic signaling complex”, which includes α2δ-1, upon binding of TSP. This complex may then nucleate a synaptic adhesion by the recruitment of cell adhesion and scaffolding proteins to the potential synaptic sites (Bolton and Eroglu, 2009). The binding of TSP to α2δ-1 may trigger a structural rearrangement in the latter, as it is known that VWF-A domains can switch a protein’s structure from an inactive to an active state upon ligand binding (Whittaker and Hynes, 2002). The activated TSP/α2δ-1 complex can then induce an intracellular signaling cascade that ultimately leads to the formation of a synapse (Fig. 1). Because gabapentin can inhibit the TSP-α2δ-1 interaction and thus inhibit TSP-induced synapse formation, it is possible that this drug arrests α2δ-1 in the inactive conformation (Fig. 1), thereby preventing formation of the complex. The inhibitory neurotransmitter GABA was also shown to inhibit TSP-2-induced synapse formation at high concentrations, raising the possibilities that GABA is the physiological ligand for α2δ-1 and that inhibitory neurotransmission in the brain may control the extent of excitatory connections (Eroglu et al., 2009). The nature of TSP-α2δ-1 downstream signaling and the constituents of the proposed synaptogenic signaling machinery are currently unknown. TSP-1 was recently found to bind to neuroligins (Xu et al., 2010), a family of postsynaptic adhesion molecules that are essential for CNS synapse function (Craig and Kang, 2007), thus it is possible that TSPs, α2δ-1 and neuroligins are components of the proposed synaptogenic signaling complex.

Figure 1. The model for a “synaptogenic signaling complex” activated by TSP that potentially regulates prosynaptogenic-signaling pathways.

(A) The α2 portion of the neuronal calcium channel subunit α2δ-1 (blue) is entirely extracellular while the δ-1 (purple) portion contains a transmembrane domain. Astrocyte-secreted TSP binds to the VWF-A domain of α2δ-1 via its EGF-like repeats (red) to cause a conformational change in the molecule from closed to open. The activated α2δ-1 then recruits an as yet unidentified signaling partner(s) (binding region unknown) to form a “synaptogenic signaling complex” which exerts multiple effects known to influence synaptogenesis through interactions with a variety of signaling molecules. TSP is involved in modulating the reorganization of the ECM by interacting with TGF-β1 as well as regulating the balance between MMPs and TIMPs. TSP influences the actin cytoskeleton via EGFRs. The synaptogenic complex may also induce clustering of various neurotransmitter receptors and scaffolding proteins via neuroligins, promoting adhesion between a presynaptic axon and postsynaptic dendrite. (B) Binding of either gabapentin or GABA keeps α2δ-1 in its inactive closed state. TSP cannot bind to α2δ-1 in this conformation, thereby preventing the downstream signaling pathways that promote synaptogenesis.

TSPs are known to control the actin cytoskeleton. Rearrangement of the actin cytoskeleton plays a pivotal role in synapse formation and remodeling as it is particularly linked to morphogenesis of the postsynaptic specializations of excitatory connections, dendritic spines (Svitkina et al., 2010). Spine number and morphology are controlled by signaling pathways initiated by regulatory Rho family GTPases, which are able to “switch on” signal transduction pathways when in a GTP-bound (i.e. active) state. Once triggered by GTPase-activating proteins (GAPs) or guanine exchange factors (GEFs), active GTPases including Rac1 and RhoA go on to induce downstream effectors including various actin polymerizing/depolymerizing factors (Hall, 1998). This pathway ultimately triggers reorganization of the actin cytoskeleton in dendrites, contributing to the initial contact and stabilization of synapses (Tashiro and Yuste, 2004). Previous studies in fibroblasts and myoblasts have shown that TSP-1 induces the formation of F-actin microspikes containing the actin-bundling protein, fascin (Adams, 1995; Adams and Schwartz, 2000). TSP-1’s ability to alter cytoskeletal organization is mediated in part by the intracellular protein muskelin (Adams et al., 1998) which acts as a trafficking protein to help move cargo along the actin cytoskeleton (Heisler et al., 2011).

The TSP-α2δ-1 interaction is crucial for TSP’s synaptogenic function; however, other previously described TSP receptors and signaling pathways are also present in neural tissues (Table 1) (Venstrom and Reichardt, 1993) and they may have additional functions at the synapse. For example, TSP-1 stimulates focal adhesion disassembly via its N-terminus binding to calreticulin [CRT]/low density lipoprotein-receptor related protein-1 [LRP-1], leading to inactivation of the small GTPase RhoA (Murphy-Ullrich, 2001; Orr et al., 2004; Sweetwyne and Murphy-Ullrich, 2012). Since RhoA signaling has been linked to synaptic plasticity (Murakoshi et al., 2011), CRT/LRP-1 signaling may thus provide a potential avenue for TSP-1 to regulate synaptic function.

Also relevant to TSP-1’s signaling capabilities are its synaptogenic EGF-like repeats, which were discovered to induce EGFR transactivation in epithelial cells (Liu et al., 2009). Specifically, activation of EGFR/ErbB2 by TSP-1 resulted in alterations in endothelial barrier function, with disruption of cell-cell junctions (Garg et al., 2011). EGFRs are also known to bind to and influence the actin cytoskeleton (Rijken et al., 1991) and play a role in neurite outgrowth (Goldshmit et al., 2004). Recent studies have also linked TSP-2 to the Notch signaling pathway, which is known to play a major role in development and homeostasis of multiple organ systems including the CNS (Meng et al., 2009; 2010).

The interactions between TSP and these various signaling molecules surely lie at the crux of CNS development and synaptogenesis. There is also a distinct possibility that these pathways are critical for another type of synaptic organization, which occurs following injury to the CNS.

3. Thrombospondin and the matricellular protein response to CNS injury

Reorganization of synaptic networks is a hallmark of many neurological disorders including but not limited to traumatic brain injury (TBI), stroke, epilepsy, and Alzheimer’s disease (AD) (Cavazos and Cross, 2006; Hu et al., 2004; Perez-Cruz et al., 2011; Rossini and Dal Forno, 2004). For example in stroke, ischemic insult results in widespread dendritic beading and spine loss that is transient when residual blood flow is available, but this damage can become permanent when energy reserves are severely compromised (Risher et al., 2010). Loss of spines, which are normally present postsynaptically at the majority of synapses in the brain (Harris and Kater, 1994), may result in alteration of function (Zhang and Murphy, 2007) or even cortical remapping (Sigler et al., 2009). Promotion of new synaptogenesis is an attractive target for stroke therapy (Chen et al., 2002; Nudo, 2003), as restoration of synaptic circuits has been shown to improve functional recovery after stroke (Adkins et al., 2008; Jones et al., 2009).

3.1 The role of thrombospondin in recovery from stroke and other brain injuries

Considering their critical roles in developmental synaptogenesis, it is perhaps not surprising to find that many matricellular proteins have been shown to be upregulated after CNS injury. Increased TSP-1 originating from brain macrophages was detected after cortical lesioning, with a reported stimulatory effect on neurite outgrowth (Chamak et al., 1994). Spinal cord injury results in a dramatic increase in TSP-1 expression (Benton et al., 2008), while TSP-1 and TSP-2 expression levels are significantly increased following stroke (Lin et al., 2003; Zhou et al., 2010), where their upregulation is primarily driven by purinergic signaling in astrocytes (Tran and Neary, 2006). Liauw and colleagues (2008) further investigated the role of TSP in stroke recovery. Occlusion of the distal middle cerebral artery and common carotid artery confirmed an increase in TSP-1 and 2 mRNA and protein levels that colocalized primarily to astrocytes. Focal ischemia in both wild type and TSP-1/2 double null (KO) mice revealed significant deficits in synaptic recovery and axonal sprouting in the TSP KO mice. These morphological deficits correlated to decreased functional recovery as assessed by the tongue protrusion test of motor function despite similar infarct size and blood vessel density between KO and wild type (Liauw et al., 2008), suggesting that TSP-1/2 deficiency did not significantly impact angiogenesis. Though these findings suggest that TSP is beneficial in terms of synapse recovery, it has been reported that inhibition of nitric oxide signaling by TSP-1 limits tissue reperfusion following stroke (Isenberg et al., 2009) and that blocking TSP-1 signaling may actually improve stroke outcome (Maxhimer et al., 2009). Further studies are needed to determine if the synaptogenic potential of TSP outweigh its negative regulatory effects in the ischemic brain.

TSP’s ability to alter synaptic remodeling after cortical injury is similarly crucial for developmental plasticity. Eroglu and colleagues (2009) used a “barrel cortex plasticity” assay to investigate the role of TSP in injury-dependent plasticity in the developing CNS. Mouse whiskers project to the brain as a topographically ordered “somatotopic” map (Erzurumlu et al., 2006), which is organized into “barrels” in the primary somatosensory cortex. Lesion of the whiskers results in structural changes in the associated barrel and its neighbours, thereby providing a direct measure of injury-dependent cortical plasticity. The investigators administered gabapentin after whisker removal to preclude the effects of TSP-induced synaptogenesis. This resulted in an atypical barrel cortex plasticity following whisker lesioning compared to saline-injected controls (Eroglu et al., 2009). This result directly implicated TSP-1/2 to be involved in cortical synaptic remodeling and plasticity after peripheral injury. Aberrant cortical remodeling also occurred following whisker lesioning in TSP-1/2 KO mice, further confirming that TSP-induced synapse formation is required for proper barrel rewiring following injury.

The actions of TSP during cortical injury may also involve the regulation of matrix metalloproteinases (MMPs). As their name implies, MMPs are capable of cleaving components of the ECM, resulting in ECM remodeling and changes in cell behavior (Ethell and Ethell, 2007). In stroke, MMPs are responsible for breaking down the basal lamina of cerebral blood vessels, resulting in disruption of the blood brain barrier (BBB) and edema formation (Montaner et al., 2001). TSP-1 induces upregulation of MMP-9 in endothelial cells (Qian et al., 1997), and MMP-9 has in turn been shown to stimulate the release of vascular endothelial growth factor (VEGF) from the ECM, triggering the so-called “angiogenic switch” (Bergers et al., 2000). Alternatively, TSP-1 can inhibit this process via suppression of MMP-9 activation, resulting in decreased VEGF secretion which contributes to TSP’s role as an inhibitor of angiogenesis (Rodriguez-Manzaneque et al., 2001). TSP-1/2 regulate MMP-2 clearance via binding to LRP1 (Emonard et al., 2004; Yang et al., 2001), while TSP-1 also regulates the expression of tissue inhibitors of metalloproteinases (TIMPs), indicating that TSP determines the precise balance of MMPs and thus influences the organization of the ECM (John et al., 2009). TSP-2 has a similar association with MMP-2, granting the TSP family tight control over the expression of these versatile proteases which can remodel the ECM according to biological needs (Agah et al., 2002; Krady et al., 2008). In the heart, MMPs mediated by TSP are crucial for myocardial remodeling after injury (Schellings et al., 2009). In the brain, MMPs have been implicated in the establishment of synaptic networks (Ethell and Ethell, 2007), as deficits in MMP-9 manifest in impairments in long-term potentiation (LTP) and behavior (Nagy et al., 2006). Thus, the timing of MMP expression regulated by TSP is critical to different biological processes, as increased induction of MMPs may be important for development while a decrease in MMP expression may represent a mechanism for TSP to reduce brain injury.

TSP may also be involved in the inflammatory response to CNS injury. Inflammation is a hallmark of nearly all neurological disorders, highlighted by the induction of microglia, monocytes, cytokines and other mediators of cell injury (Wee Yong, 2010). In the immune system, TSP-1 modulates the adhesion and aggregation of monocytes/macrophages via interactions with CD36 and CD47 (Yamauchi et al., 2002b). Accordingly, decreased inflammation and increased white matter sparing following spinal cord injury were seen when the TSP-1 receptor CD47 was knocked out, though curiously this protection did not extend to TSP-1 knockout mice (Myers et al., 2011). In addition, TSP-1 enhances the release of the inflammatory cytokines IL-6 and IL-10 via CD36 (Yamauchi et al., 2002a). There is growing evidence for cytokines playing key roles in synaptic plasticity and neurogenesis (Bilbo and Schwarz, 2009; McAfoose and Baune, 2009), such as IL-6’s importance to the modulation of LTP, a form of synaptic plasticity associated with learning and memory (Balschun et al., 2004). However, TSP-1 can also suppress IL-10 release by interacting with latent transforming growth factor beta-1 (TGF-β1) (Yamauchi et al., 2002a). Indeed, TSP-1 is known to activate latent TGF-β1 (Schultz-Cherry et al., 1994; Schultz-Cherry and Murphy-Ullrich, 1993). Given the upregulation of TGF-β1 after cortical injury (Makwana et al., 2007), it is a strong possibility that the interaction between TSP and TGF-β1 may play important roles in the injured brain. Moreover, TGF-β1 is also linked to multiple aspects of synapse formation and plasticity particularly in the peripheral nervous system (PNS) (Banyai et al., 2010; Feng and Ko, 2008; Packard et al., 2003). TGF-β1 released by Schwann cells (the major glial cell type in the PNS) promotes synapse formation at the neuromuscular junction (Feng and Ko, 2008) potentially through binding to the synaptogenic proteoglycan agrin.

3.2 Altered thrombospondin expression in neurological disease

Abnormalities in the expression of TSP and other matricellular proteins may underlie cortical dysfunction in neurological disorders that are characterized by aberrant synaptic networks. For example, altered spine morphology and reduced synaptic density in Down’s syndrome (DS), the most common genetic form of mental retardation, have recently been linked to decreased TSP expression in DS astrocytes (Garcia et al., 2010).

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder in which massive synapse loss occurs in the early clinical stage (Masliah et al., 1991; Sze et al., 1997). This synaptic loss shows high correlation with the cognitive decline presented by AD patients (Terry et al., 1991). TSP-1 expression is markedly decreased in vulnerable neuronal populations in the brains of AD patients (Buee et al., 1992), making any synaptogenically-driven repair mechanism unlikely to succeed. Synaptic disorganization is also a hallmark of epilepsy, in which aberrant synaptogenesis following seizures may actually contribute to hyperexcitability and thus worsen epileptic discharges (Cavazos and Cross, 2006). Seizure activity induces widespread alterations in the localization and synthesis of ECM molecules in a so-called ‘matrix response’ driven primarily by astrocytes (Hoffman et al., 1998). The finding that the anti-convulsant drug gabapentin inhibits TSP-induced synaptogenesis (Eroglu et al., 2009) may have direct implications for the drug’s mechanism of action, suggesting that the prevention of aberrant synaptic remodeling can be therapeutic. Future research will be necessary to determine if alterations in the expression of TSPs and their known binding partners may be beneficial in the treatment of brain injury and neurological disease.

4. Conclusions

TSPs are a family of matricellular proteins with functions in numerous organ systems (Murphy-Ullrich and Iozzo, 2012). Recent findings demonstrated a critical role for TSPs 1, 2 and 4 in synaptogenesis. Expressed primarily by astrocytes in the CNS, TSPs 1 and 2 promote CNS synaptogenesis during development and participate in synaptic repair after brain injury. Perhaps owing to these synaptogenic properties, elevated TSP expression in specific brain regions of human versus nonhuman primates has been suggested to underlie the distinct cognitive abilities of humans (Caceres et al., 2007). The list of known binding partners and functions of TSP is long and rapidly growing. Better characterization of TSP and its downstream signaling pathways should facilitate our understanding of the complex process of synapse formation.

Abbreviations

ACM

astrocyte conditioned media

AD

Alzheimer’s disease

AMPA

2-amino-3-(5-methyl-3-oxo-1,2-oxazol-4-yl)propanoic acid

CNS

central nervous system

CRT

calreticulin

DS

Down’s syndrome

ECM

extracellular matrix

EGF

epidermal growth factor

EM

electron microscopy

GABA

gamma amino butyric acid

GEF

guanine exchange factor

GAP

GTPase-activating protein

KO

knockout

LRP-1

low density lipoprotein-receptor related protein-1

LTP

long-term potentiation

MMP

matrix metalloproteinase

NMDA

N-Methyl-D-aspartate

P38MAPK

p38 mitogen-activated protein kinase

PNS

Peripheral Nervous System

RGC

retinal ganglion cell

TBI

traumatic brain injury

TGF-β1

transforming growth factor beta-1

TIMP

tissue inhibitor of metalloproteinase

TSP

thrombospondin

VEGF

vascular endothelial growth factor

VGCC

voltage-gated calcium channel

VWF-A

Von Willebrand Factor A