Volume 247, Issue 1 p. 75-84
Free Access

Neural stem cell therapy aiming at better functional recovery after spinal cord injury

Yicheng Zhu

Yicheng Zhu

Department of Stem Cell Biology and Medicine, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan

Search for more papers by this author
Naohiro Uezono

Naohiro Uezono

Department of Stem Cell Biology and Medicine, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan

Search for more papers by this author
Tetsuro Yasui

Tetsuro Yasui

Department of Stem Cell Biology and Medicine, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan

Search for more papers by this author
Kinichi Nakashima

Corresponding Author

Kinichi Nakashima

Department of Stem Cell Biology and Medicine, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan

Correspondence to: Kinichi Nakashima, Department of Stem Cell Biology and Medicine, Graduate School of Medical Sciences, Kyushu University, 3-1-1 Maidashi, Higashi-ku, Fukuoka, 812-8581, Japan. E-mail: [email protected]Search for more papers by this author
First published: 02 August 2017
Citations: 67


Injury to the spinal cord causes transection of axon fibers and neural cell death, resulting in disruption of the neural network and severe functional loss. Reconstruction of the damaged neural circuits was once considered to be hopeless as the adult mammalian central nervous system has very poor ability to regenerate. For this reason, there is currently no effective therapeutic treatment for spinal cord injury (SCI). However, with recent developments in stem cell research and cell culture technology, regenerative therapy using neural stem cell (NSC) transplantation has rapidly been developed, and this therapeutic strategy makes it possible to rebuild the destroyed neural circuits. In this review, we discuss the recent breakthroughs in NSC transplantation therapy for SCI. Developmental Dynamics 247:75–84, 2018. © 2017 Wiley Periodicals, Inc.


Together with the brain, the spinal cord composes the central nervous system (CNS) of our bodies. The spinal cord is a long cylindrical structure composed of nerves within the vertebral column, which extends from the medulla to the level of the first or second lumbar vertebra. The main function of the spinal cord is transmitting information between the brain and the body, which allows us to direct our body's voluntary muscle movements; monitor sensations of touch, pressure, temperature, and pain; and regulate autonomic functions such as digestion. Because of its irreplaceable function, injuries to the spinal cord usually cause disastrous consequences. Traumatic injury of the spinal cord disrupts the descending and ascending axonal tracts and interrupts the communication between the brain and the body, leading to the loss of motor and sensory functions that are controlled by spinal segments below the lesion level. Although the incidence of spinal cord injury (SCI) is relatively low, it often results in disastrous emotional, social and economic impacts on the injured person and their family, community, and society. To date, there is no effective cure for SCI and the only therapeutic option for SCI patients is physical rehabilitation (Behrman and Harkema, 2007; Harvey, 2016).

To reverse outcomes caused by SCI, it is necessary to repair the damaged neural circuits. However, in comparison with the peripheral nervous system (PNS), the CNS has poor regenerative ability. Once transected, CNS axons form dystrophic endbulbs at their proximal tips, termed retraction bulbs, which render neurons unable to regenerate (Bradke et al., 2012). This regeneration failure is considered to be caused by the presence of extrinsic growth inhibitory molecules including chondroitin sulfate proteoglycan (CSPG) and the deficiency of intrinsic axon growth factors such as growth associated protein 43 (GAP-43) and cyclic adenosine monophosphate (cAMP) (Strittmatter et al., 1992; Benowitz and Routtenberg, 1997; Shewan et al., 2002; Silver and Miller, 2004). Although sprouting of uninjured axons can also contribute to the recovery in an incomplete SCI, this recovery is depend on the severity of the injury (Maier and Schwab, 2006; Huebner and Strittmatter, 2009; Dell'Anno and Strittmatter, 2016).

In addition to regeneration and spouting of axonal tracts, the spared host neurons remaining at the injured site have been suggested to take part in the reconstruction of damaged neural circuits by relaying signals and to contribute to functional recovery after severe SCI. However, this spontaneous recovery is still limited, which might be due to the fact that only a few host neurons exist in the injured spinal cord (Fig. 1) (Courtine et al., 2008; Flynn et al., 2011). From this point of view, recruitment of new neurons to the injury site might be therapeutically effective for the treatment of SCI.

Details are in the caption following the image

Reconstruction of injured spinal cord by host and exogenous neurons. A: Intact spinal cord in mice. In human, direct connections between CST axons and motor neurons also exist. B: Severe SCI disrupts neuronal pathway and interrupts signals between brain and body. After injury, the surviving host neurons could rebuild the damaged neural circuits by relaying signals, which, therefore, results in limited spontaneous recovery. C: Implantation of exogenous neurons increases synaptic connection, resulting in the promotion of functional recovery.

Endogenous Neural Stem Cells in the Repair of SCI

The discovery of neural stem cells (NSCs) present in the adult brain and spinal cord provided evidence that the CNS may have the potential to repair insults by generating newborn neurons (Johansson et al., 1999; Gage, 2000). NSCs are self-renewing, multipotent cells that can give rise to neurons, astrocytes and oligodendrocytes. In the adult brain, NSCs reside in the subventricular zone and the subgranular zone of the hippocampal dentate gyrus (Bond et al., 2015). These adult NSCs continuously generate functional neurons throughout life, and this generation is suggested to be critical for biological functions such as olfaction, learning and memory (Ming and Song, 2011). In pathological conditions such as stroke and traumatic brain injury, NSCs have been shown to undergo increased proliferation and neurogenesis. Those newly generated neuronally committed cells then have the ability to migrate into the damage site and replace the damaged neurons (Zhang et al., 2004; Richardson et al., 2007).

In the spinal cord, ependymal cells lining the central canal have been revealed to possess the capacity to differentiate into neurons and glia cells in vitro, and they are defined as NSCs (Sabelstrom et al., 2014). However, in contrast to NSCs in the brain, most of the ependymal cells differentiate into astrocytes, and no neurogenesis has been observed after SCI (Namiki and Tator, 1999). Moreover, in response to injury, these generated astrocytes become reactive and eventually form a glial scar. Although a recent study provided a new concept that the glia scar may support the regeneration of CNS axons (Anderson et al., 2016), it is generally considered as a major barrier for axon regeneration (Silver and Miller, 2004).

As endogenous NSCs in the spinal cord are unable to generate neurons for neural reconstruction, introduction of exogenous neurons might seem to be a reasonable approach for SCI treatment (Fig. 1). However, due to their vulnerability and inability to proliferate, differentiated neurons are not an ideal cell source for transplantation. An alternative approach for supplying exogenous neurons is to transplant NSCs and to induce the differentiation of these donor cells toward the neuronal lineage.

Transplantation of NSCs for the Treatment of SCI

Transplanted NSCs have been suggested to exert their therapeutic effect by several supportive functions including anti-inflammation, remyelination, neuroprotection, and promotion of neurite outgrowth (Mothe and Tator, 2013). In addition to those potential functions, we and others have proved that NSCs also directly reconstruct the disrupted neural pathway by forming functional neuronal relay (Table 1).

Table 1. Summary of Recent Studies of NSCs Transplantation for SCI Focusing on Neuronal Relay
Reference Donor cells Method Results
Ogawa et al., 2002 Fetal rat spinal cord-derived NSCs NSCs were transplanted into rat spinal cord at 9 days after contusion injury at cervical level 4 (C4) Graft neurons extended processes into host tissue and formed synaptic formation with host neurons confirmed by immunostaining and electron microscopy
Cummings et al., 2005 Human fetal brain-derived NSCs NSCs were injected into NOD-SCID mice spinal cord at 9 days after contusion injury at thoracic level 9 (T9) Majority of graft NSCs differentiate into neurons and oligodendrocytes. Synaptic formation between host and graft neurons was confirmed by immunostaining and electron microscopy
Abematsu et al., 2010 Fetal mouse brain-derived NSCs NSCs were injected into the mouse spinal cord at 7 days after contusion injury at T9 level Administration of VPA promoted neuronal differentiation. Neuronal relay was confirmed by immuno-electron microscopy and WGA-tracing
Nori et al., 2011 Human iPSCs-derived NSCs NSCs were injected into NOD-SCID mice spinal cord at 9 days after contusion injury at T10 level Synaptic formation between host and graft neurons was confirmed by immunostaining and electron microscopy
Bonner et al., 2011 Fetal rat spinal cord-derived neuron-restricted precursors and glia-restricted precursors NSCs suspension was injected into the rat spinal cord immediately after dorsal column lesion at C1 level. 7 days later, BDNF-expressing lentivirus was injected into the dorsal column nuclei to guide graft axons CTβ-labeled sensory axons grew into the graft and graft extended axons into the dorsal column nuclei . Synaptic formation between host and graft neurons was confirmed by immunostaining and electron microscopy
Fujimoto et al., 2012 Human iPSC-derived NSCs NSCs were injected into NOD-SCID mice spinal cord at 7 days after contusion injury at T9 level Neuronal relay was confirmed by WGA-tracing and immunostaining
Lu et al., 2012 Fetal rat and human spinal cord-derived NSCs NSCs embedded in growth factor-containing fibrin matrix were transplanted into the rat spinal cord at 14 days (rat source) or 7 days (human source) after complete transection at T3 level Graft neurons extended axons over long distances and formed synapses with host neurons confirmed by immunostaining and electron microscopy
Hou et al., 2013 Fetal rat NSCs derived from brainstem (BS-NSCs) or spinal cord (SC-NSCs) NSCs embedded in growth factor-containing fibrin matrix were transplanted into the rat spinal cord at 14 days after complete spinal cord transection at T4 level BDA-traced host axons grew into grafts and graft-derived neurons extended axons to the intermediolateral cell column. Synaptic formation was confirmed by immunostaining
Medalha et al., 2014 Fetal rat spinal cord tissue containing neuron-restricted precursors and glia-restricted precursors Fetal rat spinal cord tissue was transplanted into rat spinal cord immediately after complete transection at T10 level Grafted cells migrated and extended processes into host tissue. BDA-traced and 5-HT positive host fibers grew into the transplants
Sharp et al., 2014 Fetal rat spinal cord-derived NSCs NSCs embedded in growth factor-containing fibrin matrix were transplanted into the rat spinal cord at 14 days after complete transection at T3 level Neuronal and glial differentiation of graft NSCs. Extensive outgrowth of axons from graft while only limited ingrowth of host axons confirmed by immunostaining
Lee et al., 2014 Fetal rat spinal cord tissue Fetal rat spinal cord tissue was transplanted into rat spinal cord immediately after cervical hemisection at C2 level Electrophysiological analyses and PRV tracing indicates the presence of host-graft interactions which has the effect on respiratory modulation
Yokota et al., 2015 Fetal mouse brain-derived NSCs NSCs were injected into mouse spinal cord immediately after contusion injury at T9 level Immunostaining and reverse transcriptase polymerase chain reaction demonstrated the expression of pre- and post-synaptic molecules in graft neurons, indicating the synaptic formation between host and graft neurons
Adler et al., 2017 Fetal mouse spinal cord-derived NSCs expressing monosynaptically restricted reporter viruses NSCs were injected into mouse spinal cord immediately after dorsal column lesion at C4 level Immunostaining indicates graft neurons receive synaptic inputs from all host spinal cord tracts
  • NOD-SCID, non-obese diabetic, severe combined immune deficiency.

Early transplant works performed in the 1980s used fetal spinal cord grafts for SCI and indicated that some axonal projections could be established between host and donor tissues, first providing evidence of exogenous neuronal relay (Reier et al., 1986). Then, several studies have described transplantation of rodent NSCs results in partial recovery and this recovery is due to the neuronal differentiation of transplanted NSCs (Ogawa et al., 2002; Lee et al., 2014). However, in severe SCI, only a small fraction of exogenous NSCs differentiate into neurons, while most of them differentiate into astrocytes (Setoguchi et al., 2004; Abematsu et al., 2006; Martino and Pluchino, 2006). This astrocytic differentiation is considered to be induced by the microenvironment of the injured site, including inflammatory cytokines released following SCI such as interleukin-6 (IL-6) and tumor necrosis factor-alpha (TNF-α) (Bonni et al., 1997; Nakamura et al., 2003; Abematsu et al., 2006; Ricci-Vitiani et al., 2006).

To avoid this undesirable differentiation driven by the extracellular environment and to yield more functional neurons, we have previously focused on the intracellular program, namely the epigenetic mechanism, and we showed that manipulating the epigenetic status by histone deacetylases (HDACs) inhibitor promoted the differentiation of transplanted NSCs into neurons rather than glial cells (Abematsu et al., 2010). In that study, we found that mice receiving combination therapy with an HDAC inhibitor valproic acid (VPA, also known as an antiepileptic drug) administration and NSC transplantation, which is named the HINT (HDAC Inhibitor and NSC Transplantation) method, exhibited marked functional recovery compared with control mice. Furthermore, immunostaining of the spinal cord revealed that VPA treatment suppresses astrocytic differentiation and effectively promotes neuronal differentiation of transplanted NSCs (Fig. 2).

Details are in the caption following the image

VPA-induced neuronal differentiation and the HINT method. A: After transplantation into the injury site, most NSCs differentiate into astrocytes due to the effect of extracellular factors secreted around the lesion area. B: VPA administration inhibits astrocytic differentiation and promotes neuronal differentiation of transplanted NSCs. C: Treatment with VPA (HDAC inhibitor) enhances histone acetylation and increases expression of the basic helix-loop-helix transcription factor NeuroD, which, therefore, promotes neuronal differentiation of NSCs.

To successfully relay the neuronal signals, the injured axonal tracts or host interneurons must make synapses with transplant-derived (graft) neurons with an appropriate phenotype, and the graft neurons must extent their axons to connect with the host targets (Bonner and Steward, 2015). To confirm the connection between host and graft neurons, many studies were performed for morphological analysis using electron microscopy and the host-graft-host synapse formation was revealed (Bonner et al., 2011; Nori et al., 2011). By applying laser microdissection, Yokota and colleagues reported increased transcription level of pre- and postsynaptic molecules in graft neurons, which suggests graft neurons may both receive and transmit synaptic signals between host neurons (Yokota et al., 2015).

In addition to these morphological and quantitative polymerase chain reaction analysis, application of trans-synaptic tracer such as wheat germ agglutinin (WGA) and pseudorabies virus (PRV) not only presents evidence that the synaptic formation exists between host and graft neurons, but also indicates that the novel neural circuit is functional (Abematsu et al., 2010; Fujimoto et al., 2012; Lee et al., 2014). In the study published in 2010, we injected WGA into the motor cortex of the mouse brain and found that WGA was delivered from the primary motor neurons in the cerebral cortex to neurons derived from transplanted NSCs, and further transported to the ventral horn neurons residing in the caudal region below the injured site, demonstrating the presence of active synaptic connection between host and graft neurons (Abematsu et al., 2010). However, it is still unclear which neural circuit the graft neurons are incorporated into. To solve this problem, a group recently designed special graft NSCs that express monosynaptically restricted reporter viruses. In this system, because fluorescently labeled, any host neurons that make monosynaptic contact to graft neurons could be detected and their results indicated that graft neurons receive numerous host inputs that normally innervates the spinal cord (Adler et al., 2017).

Toward Clinical Application of NSC Transplantation

Several studies showed that transplantation of NSCs derived from human fetal tissues or embryonic stem cells (ESCs) promoted functional recovery in animal SCI models (Cummings et al., 2005; Iwanami et al., 2005; Keirstead et al., 2005; Sharp et al., 2010). However, regarding clinical applications and ethical issues, the derivation and use of these cells is highly controversial (Carvalho et al., 2013; de Miguel Beriain, 2014). One of the most exciting breakthroughs in regenerative medicine was the establishment of induced pluripotent stem cells (iPSCs) (Takahashi and Yamanaka, 2006). In 2007, Takahashi et al. and Yu et al. almost simultaneously reported that iPSCs can also be generated from adult human fibroblasts (Takahashi et al., 2007; Yu et al., 2007). The use of iPSCs derived from human somatic cells can, therefore, avoid ethical problems as it does not necessitate the use of human embryos, and can solve the immunological rejection problem as iPSCs can be created from the tissue of SCI patients who will receive the transplantation. Based on iPSC technology, cell replacement therapy has been greatly advanced toward clinical application.

By generating NSCs from human iPSCs (hiPSCs), we and others have shown that transplantation of hiPSC-derived NSCs improved functional recovery in SCI model animals (Nori et al., 2011; Fujimoto et al., 2012; Kobayashi et al., 2012). Moreover, in contrast to rodent NSCs, the vast majority of hiPSC-derived NSCs differentiated into neurons after transplantation in the spinal cord (Nori et al., 2011; Falk et al., 2012; Fujimoto et al., 2012). Consistent with previous reports (Ogawa et al., 2002; Abematsu et al., 2010), neurons derived from transplanted hiPSC-derived NSCs were also found to reconstruct the damaged neural circuits by making synapses with endogenous neurons. Therefore, transplantation of hiPSC-derived NSCs represents a promising approach for the treatment of SCI.

For clinical application, it is possible to autologously transplant iPSC-derived cells from patients' own somatic cells to avoid immunological rejection reactions. However, it takes at least 4 months to prepare iPSCs, whereas the most appropriate time for SCI transplantation is considered to be 2 to 4 weeks after injury, making it difficult to perform autologous transplantation (Nakamura et al., 2003; Takahashi et al., 2007). Therefore, the Center for iPS Cell Research and Application (CiRA) at Kyoto University has undertaken the building of a stock of iPSCs for regenerative medicine. This project aims to collect safe iPSC lines with homozygous human leukocyte antigens (HLAs) that are associated with decreased incidence of immunological rejection (HLA-A, HLA-B, HLA-DR). In addition to this project in Japan, the United States and the United Kingdom have initiated projects by generating iPSCs prepared from HLA homozygous donors, and other countries are in planning phases of the same projects (Azuma and Yamanaka, 2016). Preparation of iPSC-derived donor cells that do not cause safety and immunological rejection problems is thus expected to reduce the time and cost of generation of transplanted cells.

Although the reprogramming technology for creating iPSCs seems effective and useful, as reprogramming-related genes are randomly inserted into DNA by viral vectors, it could potentially cause genetic mutations and consequently teratoma formation after transplantation. As a way to avoid this problem, generation of integration-free iPSCs without a viral vector has recently been reported (Okita et al., 2008, 2011; Kaji et al., 2009). In addition, methods for direct reprogramming of somatic cells into NSCs or neurons have been intensively studied and developed rapidly (Vierbuchen et al., 2010; Pang et al., 2011; Han et al., 2012; Ring et al., 2012). Generation of the objective cells without undergoing the pluripotency stage could shorten the preparation time and theoretically decrease the risk of teratoma formation. Therefore, neurons or NSCs reprogrammed directly from somatic cells have become promising candidates for nerve regenerative medicine.

Reprogramming is considered to be the result of epigenome reconstruction through changes in the chromatin structure and state. However, the detailed mechanism underlying reprogramming still remains elusive. Before reprogramming technology is used in clinical application, its mechanism should be further elucidated to eliminate the potential risks and ensure the safety for regenerative medicine.

Combination Therapy for SCI

Although transplantation of NSCs holds great promise for SCI treatment, functional recovery after this strategy is still not complete. To reconstruct the damaged neural circuits more functionally, it is necessary to combine cell replacement therapy with other therapeutic strategies (cocktail treatment) such as preserving the microenvironment or preventing neurite degeneration.

Immunoregulation Therapy

One of the candidates is combining NSC transplantation with immunoregulation therapy to prevent the expansion of tissue injury by suppressing immune responses and, therefore, preserve the microenvironment of the injured spinal cord to promote the therapeutic effect of cell transplantation.

Damage to the adult mammal CNS causes a local immune response that results in the subsequent expansion of secondary injuries such as inflammation, necrosis and apoptosis in the acute phase, leading to the collapse of neural networks. Recent studies regulating functions of cytokines related to secondary injury such as the TNF family and IL family have shown beneficial effects in SCI animal models, offering promising alternatives for clinical application in SCI.

In mouse models of SCI, TNF-α-expressing cells were initially observed around the injured site at 30 to 45 min after injury, and the expression of TNF-α and IL-6 was strongly increased from 3 up to 24 hr (Habgood et al., 2007; Pineau and Lacroix, 2007). These inflammatory cytokines are released from damaged cells, and reactive microglia then recruit immune cells from broken blood vessels and promote tissue repair. Because reactive microglia and recruited macrophages are the main inflammatory cytokine-producing cells, the inflammatory response is then further exacerbated. In addition to its proinflammatory effects, TNF-α was reported to induce neuronal apoptosis and/or necrosis by means of activating caspase 8 through the TNF receptor, and IL-1β was reported to exacerbate CNS edema by promoting the transcription of aquaporin 4, which results in blood flow disturbance and the subsequent necrosis (Ito et al., 2006; Wang et al., 2008). Administration of IL-6, which is an important trigger for induction of inflammatory cytokine expression, to the injured spinal cord causes increased neutrophils, expansion of the areas occupied by macrophages and activated microglia, and reduced axonal regeneration (Lacroix et al., 2002; Fu and Saporta, 2005). These findings suggest that inflammatory cytokines are strongly associated with the secondary damage after SCI.

Therefore, the use of antibodies against inflammatory cytokines or their cognate receptors to minimize secondary damage is a possible strategy for SCI. Okada et al. reported that IL-6 receptor monoclonal antibody suppressed the astrocytic differentiation of endogenous NSCs and decreased the number of inflammatory cells and scar formation in mouse SCI models (Okada et al., 2004). Chen et al. reported that administration of TNF-α antagonist (Etanercept) reduced the inflammatory tissue damage and improved hindlimb motor function in rat SCI models (Chen et al., 2011). However, the functional recovery gained after anti-cytokine antibody therapy alone is still far from complete.

A trial of combination therapy with TNF-α antagonist and NSC transplantation was then performed in a rat SCI model, and it was reported that rats that received combinatorial treatment exhibited better improvement in hindlimb movement than those that received TNF-α antagonist or NSC transplantation alone (Wang et al., 2014). In that study, the survival of transplanted NSCs was improved by suppression of TNF-α, and increased myelinated axons were observed after injury, suggesting that immunoregulation therapy could inhibit the expansion of secondary damage and thereby improve the NSC transplantation-mediated functional recovery. As immune reactions have both beneficial and harmful effects for living bodies, the potential risks should be taken into consideration before combining the immunoregulation therapy with NSC transplantation for future clinical application.

Inhibition of Extracellular Cue-Induced Axon Degeneration

One of the possible reasons why the therapeutic effect of NSC transplantation is limited is that the integration of grafted cells into neural circuits is inefficient due to the degenerated axonal tracts. CSPG produced by scar-forming cells was reported to destroy the cytoskeleton of the axonal growth cone and thus induce neurite retraction through activation of RhoA (Silver and Miller, 2004; Yiu and He, 2006). In addition to CSPG, myelin-associated proteins including Nogo-A (Nogo-66), MAG (myelin-associated glycoprotein), and Omgp (oligodendrocyte myelin glycoprotein) and Sema3A (semaphorin 3A) present in scar tissue are also known to suppress axon elongation through RhoA signal activation (McKerracher et al., 1994; Habib et al., 1998; Prinjha et al., 2000; Tang et al., 2004) (Fig. 3).

Details are in the caption following the image

Inhibition of axon regrowth by surrounding microenvironment. CSPG produced by glial scar induces RhoA activation through leukocyte common antigen-related phosphatase (LAR) and protein tyrosine phosphatase σ (PTPσ). The myelin associated proteins (MAG, Nogo-A, Omgp) released from sheathed axons after injury bind to a common Nogo receptor (NgR) and induce RhoA activation by means of co-receptor p75NTR. Sema3A secreted from fibrotic scar could also trigger RhoA activation after binding to Neuropilin-1/Plexin-A. Activation of RhoA signaling ultimately induces growth cone collapse and results in neurite retraction.

To reconstruct the damaged neural circuits, theoretically, it would be effective to promote axon regeneration and preserve the microenvironment in combination with NSC transplantation. It was reported that combinatorial therapy of NSC transplantation and chondroitinase-ABC (ChABC, which is an enzyme with the ability to degrade CSPG) enhanced the extension of corticospinal tract (CST) axons and improved the integration of transplanted cells into the injured spinal cord, which eventually promoted functional recovery in chronic SCI rats (Karimi-Abdolrezaee et al., 2010).

Searching for Novel Transplantation Method

The most common approach for NSC transplantation in mouse models is to inject NSC-containing medium into the injury epicenter. However, as the tissue environment of the injured CNS is unfavorable for graft cell survival, only approximately 20% of transplanted cells could survive for a long period after being transplanted into the injured spinal cord (Okada et al., 2005; Abematsu et al., 2010; Fujimoto et al., 2012; Yokota et al., 2015). A study by Lu et al. showed that transplantation of fibrin matrices containing NSCs and growth factors dramatically enhanced graft survival in a rat SCI model (Lu et al., 2012). The graft-derived neurons thus obtained extended long-distance axons and restored disrupted neural circuits. Another group used a tissue engineering neural network constituted in vitro by co-culture of gene-modified Schwann cells and NSCs in a gelatin sponge scaffold and found that after transplantation into injured rat spinal cord, the constructed neural network could integrate into the host functional neural network, and even promoted axonal regeneration of host neurons (Lai et al., 2016). These studies provided evidence that co-implantation of biomaterials could create a favorable microenvironment for transplanted cells and represented a novel approach for future cell replacement medicine.

Combination With Training to Improve Functional Relay

Reconstruction of a functional neural circuit depends on specific connections between host and graft neurons. In an intact spinal cord, axonal tracts make synapse with appropriate target neurons and this connection is formed in early development through several mechanisms, including activity-dependent plasticity (Martin et al., 1999; Tahayori and Koceja, 2012). Different from the normal neural circuit in the brain and the spinal cord, graft neurons randomly connect with host neurons. In theory, only if the graft neuron bridges between damage axon and its normal target, the functional recovery could be achieved. Therefore, strategies that facilitate establishment of specific connections between graft and host neurons should be a considerable combinatorial approach for NSCs transplantation.

In 2014, Lee and colleagues showed that hypoxic conditions after transplantation of fetal spinal cord tissue into the phrenic nucleus increased patterned bursting around graft neurons, suggesting that posttransplantation training such as hypoxia may potentially improve the functional connectivity between graft and host neurons (Lee et al., 2014). Another recent study reported that even in chronic SCI, the functional recovery could be enhanced by combination therapy with NSCs transplantation and treadmill training (Tashiro et al., 2016). Although the underlying mechanism that rehabilitation facilitates synaptic connectivity remains unclear, these studies indicate that training may be a useful tool to enhance the host-graft integration and achieve better functional recovery.

Future Directions

Because of their abilities to self-renew and differentiate into distinct cell types (i.e., neurons, astrocytes, and oligodendrocytes), NSCs hold great promise for cell replacement therapy in neurological diseases such as Parkinson's disease, stroke, multiple sclerosis, and SCI. Transplantation of NSCs has been reported to improve functional recovery after SCI. While several studies suggested that transplanted NSCs contribute to the recovery by their supportive role, e.g., secreting various types of factors to suppress myelin inhibitors, prevent neuronal cell death, enhance remyelination, and promote axon regeneration (Mothe and Tator, 2013), we and others proved that besides those functions, implanted cells also integrate into the damaged neural circuits (Abematsu et al., 2010; Nori et al., 2011; Fujimoto et al., 2012). Neurons derived from transplanted NSCs make synapses with disrupted CST axons and endogenous neurons in the ventral horn, thereby allowing neural signals from the brain to pass through the injured cord and improve motor function recovery (Abematsu et al., 2010; Fujimoto et al., 2012).

To reconstruct a functional neural circuit, the subtypes of transplant-derived neuron should be taken into consideration. As most of axonal tracts are glutamatergic, introduction of glutamatergic donor neurons may be theoretically beneficial for functional relay (Bonner and Steward, 2015). In a recent study, researchers have succeeded in generating glutamatergic V2a interneurons from human iPSCs and transplanting such cells into the mouse spinal cord. Immunostaining data show transplanted V2a interneurons formed synapses with host neurons. As glutamatergic V2a interneurons are critical for respiration and locomotion, these cells would thus be an ideal source for SCI transplantation (Butts et al., 2017).

In addition to the neuronal phenotype, a recent study have reported that after transplantation of caudalized but not rostralized neuronal grafts, damaged CST could regenerate beyond lesion sites (Kadoya et al., 2016), which suggests that neurons derived from different regions have different effects for recovery. As transplantation therapy has no effect in severe SCI where only very few host neurons exist around lesion site (Yokota et al., 2015), the characteristic of transplant-derived neurons is, therefore, different from that of host interneurons. Further studies are needed to analyze the characteristic of host interneurons, and transplantation of neurons with similar characteristic to host interneurons might be beneficial for future SCI therapy.

In addition to recruiting exogenous neurons to reconstruct the damaged neural circuits, conversion of endogenous cells around the lesion site to another cell lineage that is beneficial for injury site restoration has also become a possible replacement therapy under investigation in current regenerative medicine research.

The first study showing that neuronal conversion could be achieved in vivo was reported in 2013. In that study, Torper et al. introduced three reprogramming genes, Ascl1, Brn2, and Myt1l, to the astrocytes in the mouse striatum and converted these astrocytes into neurons (Torper et al., 2013). Another study published in the same year showed that injection of lentiviral vector carrying a gene for a single factor, Sox2, into mouse brain is sufficient to reprogram astrocytes into neuroblasts which subsequently develop into functionally mature neurons (Niu et al., 2013). In 2014, the same group proved that expression of Sox2 could also induce the conversion of astrocytes to neurons in both intact and injured mouse spinal cords (Su et al., 2014). However, the conversion efficiency and the number of converted neurons were low, restricting the functional recovery.

Another transcription factor, NeuroD1, has been shown to be more effective for converting astrocytes into neurons in vivo (Guo et al., 2014). In that study, astrocytes in the injured mouse brain were infected with retrovirus expressing NeuroD1, and it was found that the majority of NeuroD1-expressing astrocytes were reprogrammed into functional glutamatergic neurons. In addition, oligodendrocyte precursor cells could be reprogrammed into both glutamatergic and GABAergic neurons after being forced to express NeuroD1. The reprogramming efficiency achieved by using NeuroD1 is almost 90%, making it a potential candidate for neuronal transition. Considering that the glial scar which prevents axon regrowth after SCI consists predominantly of reactive astrocytes, replacement of injury-induced reactive astrocytes with functional neurons could be an alternative therapeutic strategy for SCI (Fig. 4).

Details are in the caption following the image

In vivo direct reprogramming for the treatment of SCI. Schematic representation of in vivo reprogramming strategy for SCI. Glial scar consisting of astrocytes represents a major barrier for axonal regeneration. Therefore, reprogramming of these scar-forming astrocytes into functional neurons has a potential therapeutic effect for SCI. Reprogrammed neurons could theoretically reconstruct damaged neural circuits by relaying signals.

With the recent development of three-dimensional (3D) culture technology, it is already possible to generate and culture cerebral organoids from human pluripotent stem cells (ESCs and iPSCs), which makes it possible to investigate the pathogenesis of various CNS disorders (Kelava and Lancaster, 2016). Although generation of spinal cord organoids has not been reported yet, Gingras et al. developed an in vitro 3D model of motor neuron production by using cells from mouse spinal cord, providing a platform for studying motor neurite outgrowth (Gingras et al., 2008). In the future, it is expected that an in vitro organoid model of the spinal cord would help researchers to clarify the pathophysiology of SCI in more detail and develop treatment methods without using model animals.


We thank Dr. T. Matsuda and Dr. K. Irie for providing original illustrations.