Mechanical forces usually cause direct damage to the normal tissue and disrupt local homeostasis when patients or animals undergo SCI, which on the other hand triggers multitudinous multi-cellular responses. Although it is incompletely understood how mechanical forces and damaged tissues initially trigger the activation of astrocytes after SCI. The previous study has identified that astrocytes are susceptible to membrane distortions and debris (75). Traumatic membrane deformation could activate mechanosensitive ion channels and result in the rapid influx of extracellular calcium and sodium in astrocytes (76–78). Other studies show that plasma membrane stretching can rise the release of intracellular calcium and ATP via extracellular signal-regulated protein kinase (ERK) and PKB/Akt signaling pathways (79, 80). Besides, astrocytes may also release endothelin-1 (ET-1), isoprostanes, and matrix metalloproteinases 9 (MMP-9) after stretch-induced injury (81, 82). More studies are needed to have a deeper understanding of these.

Accumulating studies have identified that a lot of molecules, such as chemokines, cytokines, transcription factors, and growth factors are the mediators for the activation of astrocytes (factors are shown in Table 2). For example, proinflammatory cytokines, such as TNF-α, interleukin (IL)-6, and IL-1β initially trigger the reactivity of astrocytes during the acute phase after SCI while other molecules maintain astrocytes reactivity in the later stages (83–85). Additionally, it is worth mentioning that reactive astrocytes can release triggering molecules, such as TNF-α, IL-6, and MMP-9, which in turn activate more astrocytes (86). Besides, other glial cells, such as activated microglia, are identified to induce the activation of astrocytes by secreting various factors, such as Il-1α, TNF, and C1q (87). Other related molecules involved in the activation of astrocytes are shown in Table 2.

Many signaling pathways are closely involved in the activation of astrocytes, such as STAT3, TGF-β, NF-κB, JNK/c-Jun, and MAPK (more signaling ways are shown in Table 2). Here we will mainly introduce TGF-β and STAT3 signaling pathways. The STAT3 signaling pathway is one of the most important signaling pathways to mediate the formation of reactive astrocytes. Mice with STAT3 knock-out in astrocytes showed the attenuated upregulation of GFAP, unsuccessful cell hypertrophy, and failed scar formation after SCI (88). Other groups also identified that selective STAT3 deletion in mice could limit the migration of astrocytes and result in the widespread infiltration of inflammatory cells, degeneration of neurons, and demyelination of axons that can lead to severe motor deficits. However, by the activation of STAT3 signaling pathway, they observed that reactive astrocytes migrated rapidly around the lesion and secluded inflammatory cells that lead to a notable improvement in functional recovery (89). These results provided a potential intervention target of STAT3 signaling pathway in the treatment of SCI. TGF-β signaling pathway greatly contributes to the formation of reactive astrocytes. As discussed above, TGF-β is a key upstream trigger in the formation of reactive astrocytes. The previous study has shown that TGF-β could increase the expression of anti-regenerative molecules, such as CSPGs, laminin, and fibronectin by several-fold in reactive astrocytes (90). Interestingly, fibrinogen could act as a stimulating factor which can activate TGF-β signaling pathway, as a result, it could induce the activation of astrocytes and formation of CSPGs (91). In addition, it could induce astrogliosis by injecting fibrinogen into the mouse cortex (91). On contrary, with the genetical ablation of fibrinogen in mice, they found inhibited TGF-β activation and hampered glial scar formation (91). Other signaling pathways are shown in Table 2.

Recent years, numerous studies have identified that the activation of astrocytes could lead to the change of functions with releasing a range of molecules, such as cytokines [TNF-α, IL-6, IL-10, IL-1β, etc. (85, 92, 93)], chemokines [CCL2, CCL3, etc. (94, 95)], growth factors [BDNF, GDNF, etc. (96, 97)], toxic amino acids [GABA and glutamate (98, 99)], extracellular matrix [CSPGs, collagen I, fibronectin, MMP-9, etc. (100–102)], and intermediate filaments [Nestin, vimentin, and GFAP (100, 102)], which would have a significant influence on the spinal cord microenvironment after SCI (Table 2). The molecules released by reactive astrocytes can activate more normal astrocytes into reactive astrocytes and contribute to glial scar forming. On the other hand, they also affect other cells, such as neurons, OPCs, and microglia through a variety of complexed effects (71).

Over the past years, reactive astrocytes were thought to be detrimental for recovery after SCI. However, recent studies have identified that reactive astrocytes also contribute to SCI repair. Here, we will discuss the beneficial and detrimental effects of reactive astrocytes after SCI (Table 3).

Reactive astrocytes are considered to be a defense mechanism of astrocytes responding to SCI. After SCI, BBB breaks down and becomes leaky to endogenous and exogenous blood-borne macromolecules that can result in disastrous consequence. These changes will mediate reactive astrocytes to upregulate Sonic hedgehog (SHH) and activate signaling cascades to repair the tight junctions of the BBB (103). Interestingly, with the absence of reactive astrocytes, it was failure in repairing the damaged BBB (104). At acute stage after SCI, reactive astrocytes migrate rapidly around the lesion to seclude inflammatory cells and limit the extent of inflammation that has a notable improvement in functional recovery (89). Further, Jill et al. found significantly increased and prolonged infiltration of inflammatory cells around the lesion with selective and conditional reactive astrocytes ablation in mice (104). Various endogenous and exogenous factors result in the release and accumulation of cell debris and neurotoxic factors in the extracellular spaces after SCI. Recently, reactive astrocytes were identified to play a crucial role in removing these cell debris and neurotoxic factors. More importantly, reactive astrocytes have the ability to phagocytose dead cells in vitro and in vivo via the upregulation of ABCA1 (105, 106). Reactive astrocytes can also reduce the impact of glutamate excitotoxicity on neurons and OPCs by clearing excess glutamate from the blood or necrotic neuronal cell death (107). Besides, reactive astrocytes can affect immune cells through releasing various molecules, such as TNF-α, TGF-β, and proteoglycans. CSPGs have a close relationship with immune activity as they can recruit chemokines and growth factors that enhances the connection of immune cells (71).

Glial scar formation has been recognized for many years. After SCI, inflammatory cells (macrophages, neutrophils, and lymphocytes), fibrotic cells, and other cells, such as pericytes, fibroblasts, and OPCs migrate rapidly into the lesion, and subsequently newly proliferated, elongated reactive astrocytes come around the lesion to form a border which could separate necrotic tissue from healthy tissue (108–110). The border formed by reactive astrocytes can limit further expansion of the lesion and restrict inflammatory cells within damaged tissue that will protect the surrounding viable neural tissue from secondary damage (111). Further, selective inhibition of astrocyte reactivity results in the widespread propagation of inflammatory cells beyond the lesion.

In addition to the above protective effects, reactive astrocytes also have detrimental effects. As discussed above, reactive astrocytes can form a physical barrier to confine the lesion, however, it can also obstruct axonal growth. Besides, reactive astrocytes secrete inhibitory proteins, such as CSPGs, which are considered to be the major inhibitors of axonal regeneration. CSPGs derived from reactive astrocytes inhibit the growth of axons in vitro, and axonal regeneration is observed to stop at CSPG-rich regions in vivo. On the contrary, Chondroitinase ABC, by removing CSPG glycosaminoglycan (GAG) chains, attenuates the inhibitory activity of CSPGs, which is shown to facilitate axonal regeneration and functional recovery (112). Further, Hyunjung Lee et al. found that using thermostabilized Chondroitinase ABC through a hydrogel-microtube scaffold system could enhance the axonal regrowth, sprouting, and improve functional recovery after SCI (113). Additionally, other studies have shown that CSPGs inhibited axonal regeneration while the inhibition of CSPGs could improve functional recovery (114).

Reactive astrocytes play a modulatory role in NPCs and OPCs post-SCI. OPCs have extremely powerful ability in remyelination as they can proliferate and differentiate into OLs that will replenish a large number of lost OLs after SCI. Recently, Justin R Siebert et al. have discovered that astrocytes-derived CSPGs highly inhibited the migration and differentiation of OPCs in vitro, and the number of OPCs surrounding the lesion significantly increased when treated with the enzyme chondroitinase ABC (115). Other study also proved that CSPGs had a dampening effect on the outgrowth and differentiation of OPCs, and treated with chondroitinase ABC could completely eliminate this inhibition (116). In addition to CSPGs, other molecules, such as BMP and ET-1 released by reactive astrocytes can also inhibit the differentiation of OPCs and finally influence remyelination (117, 118). Besides, reactive astrocytes have a role in inhibiting the neuronal differentiation of NPCs by expressing insulin-like growth factor binding protein 6 (IGFBP6) and CSPGs (119).

Microglia/macrophages phenotypes are mainly determined by the focal lesion and new stimuli can change the phenotypes. It is now well-acknowledged that microglia/macrophages are activated into different functional phenotypes after SCI. M1/M2 dichotomy is the earliest and simplest concept. M1 macrophages (or ‘classically' activated macrophages) are activated by the prototypical T helper 1 cytokine (TH1), interferon-γ (IFNγ), and lipopolysaccharide (LPS), which typically release inflammatory cytokines (IL-1, IL-6, TNFα, etc.), chemokines (CCL8, CCL 15, CXCL 10, CXCL 11, etc.), and the high levels of oxidative metabolites (ROS and NOS). On the contrast, M2 macrophages (or “alternatively” activated macrophages) are activated by the prototypic TH2 cytokine IL-4 and IL-13, which can produce numerous protective factors (TGFβ, IL-10, IL-1Ra, etc.) and clear cellular debris (120–122). However, the status and functional phenotypes of microglia/macrophages are much more complicated in vivo. Accumulating studies have identified the multiformity in M2 phenotype subpopulations, such as M2a, M2b, and M2c phenotypes, each phenotype is characterized by unique physiological features and distinct biological functions (121). Nowadays, microglia/macrophages in many other situations did not show a clear M1 or M2 phenotype or showed phenotypic plasticity during the disease progression. Single cell techniques and other new tools are, contributing to the understanding of polarization heterogeneity (123). By single-cell analysis, Lindsay M Milich at el. identified four microglial subtypes in the injured mouse spinal cord, which were labeled homeostatic, inflammatory, dividing, and migrating microglia. Homeostatic microglia were identified by several annotated markers of steady-state microglia, such as P2ry12, Siglech, and Tmem119. Inflammatory microglia were identified by the low expression of purinergic receptor P2ry12 and increased expression of Igf1. Dividing microglia expressed low levels of P2ry12, increased expression of Msr1, and high levels of cell cycle–related genes, such as Cdk1. Migrating microglia had the low levels of P2ry12, and the high levels of Msr1 and the growth factor Igf1 (124). Besides, two macrophage subtypes were named chemotaxis-inducing macrophages and inflammatory macrophages in addition to the border-associated macrophages based on their gene ontology terms. Both subtypes expressed the lysosomal gene Cd63, however, chemotaxis-inducing macrophages preferentially express heme oxygenase Hmox1 while inflammatory macrophages express Apoe (124, 125).

Activated microglia could release a large number of pro-inflammatory cytokines, chemokines, and other cytotoxic factors after SCI. They respond to SCI within minutes by producing pro-inflammatory molecules which can lead to the influx of multiple inflammatory cells from the circulation. Neutrophils are the first circulating leukocytes to infiltrate into the lesion and are prominently located in severely damaged site (126, 127). Besides, peripheral macrophages will infiltrate into the lesion and help clear apoptotic cells (127). However, these neutrophils and macrophages may be destructive to the lesion as they can produce various molecules, such as MMP-9 and disrupt the functions of the BSB (128). Besides, T and B lymphocytes are found to infiltrate into the injured lesion and cause a systemic autoimmune response (129). Here, we will mainly discuss the harmful and beneficial effects of neuroinflammation induced by activated microglia/macrophages after SCI.

Activated M1 microglia/macrophages induce neurons death and contribute to the secondary damage by releasing pro-inflammatory factors, such as IL-1β, IL-6, TNF-α, CCL5, and iNOS. Here we mainly elaborate IL-1β and TNF-α that play a detrimental role after SCI. IL-1β expressed by astrocytes and microglia was detected to reach peak at 12 h after SCI in rodents (130). IL-1β and TNF-α were proved to involve in the recruitment and activation of peripheral immune cells and the activation of astrocytes and microglia. In rats, the infusion of IL-1β markedly enhanced the cortical neuronal loss, on the contrast, it could significantly inhibit neuronal damage by IL-1 receptor antagonist (IL-1ra) (131). Other study also identified that IL-1β contributed to ischemic brain damage while IL-1ra markedly protected the focal cerebral from ischemia in the rat (132). TNF-α, another proinflammatory cytokine, expressed mainly by activated microglia/macrophages, contributes to neuronal cells death after SCI by binding to TNFRI and TNFRII (133). In addition, soluble TNFRI, which can compete with TNF-α by binding to TNFR, eventually reduces the neuronal cells death (133). Tiziana Genovese et al. indicated that the genetic inhibition of TNF-α significantly reduced the degree of inflammation, tissue injury, and apoptosis in an experimental model of spinal cord trauma (134). Besides, overexpressing TNF-α was shown to mediate OLs, OPCs death, and myelin vacuolization which could finally result in spontaneous demyelination (135).

Activated M2 microglia/macrophages have anti-inflammatory and neuroprotective effects by increasing the expression of anti-inflammatory molecules, such as IL-10, TGF-β, IGF-1, and BNDF. For example, IL-10 shows a wide range of regulatory activities in response to SCI. Tiziana Genovese et al. found that there was a significant augmentation of TNF-α, IL-1β and S100β which worsened the recovery of limb function in IL-10 KO mice (136). Recently, the group of Jessica Y Chen delivered IL-10 into mice SCI model by loading an implantable biomaterial scaffold. They observed that IL-10 could significantly reduce damage to tissue and improve subsequent motor deficits (137). IGF-I is a potent neurotrophic factor released by activated microglia/macrophages with anti-inflammatory response. The previous study showed that IGF-I gene transfer after SCI could inhibit the loss of neurons and significantly improve the neurological dysfunction (138). Besides, other study showed that BDNF and IGF-I could significantly enhance neuroprotective effects, such as repairing BSCB damage, alleviating edema, and cells injury by the downregulation of nNOS after SCI in rat model (139).

In addition to the immediate trauma damage, there is a prolonged secondary damage after SCI. OLs are quite susceptible to changes in the surrounding microenvironment after SCI which can result in the necrosis, apoptosis, and autophagy of OLs (140–142). Acute OLs death has previously been investigated to occur within 15 min after injury and the number of OLs steadily declined by 7 days post-injury (143, 144). Previous studies have identified several aspects of subsequent damage that can lead to the death of OLs. Ischemia is an apparent reason to result in OLs death in the damaged areas of white matter (145). Ischemia and reperfusion contribute to the formation of free radical, such as reactive oxygen and nitrogen species, and OLs are particularly vulnerable to the oxidative stress. After SCI, ROS (hydroxyl radicals and superoxide) and RNS (nitric oxide, peroxynitrite, and nitrated protein) were detected to be at the increased levels (146–148). By oxidizing protein, lipids and nuclear material, ROS and NOS damage OLs which results in the necrosis and apoptosis of OLs. Besides, excitotoxicity is another major factor leading to the OLs death after SCI. The glutamate will reach a toxic level after SCI that can lead to the OLs death in vitro and in vivo. Glutamate binding to AMPA/kainate glutamate receptors expressed in OLs leads to OLs death via receptor overactivation and the specific inhibitors of AMPA receptors can block OLs death (149). As discussed above, extracellular ATP released by multiple cell types after SCI can also contribute to OLs death. ATP is proved to cause OLs death via an activation of calcium-permeable P2X(7) and treatment with P2X(7) antagonists reduces demyelination and improve neurological symptoms (150). In addition, recent studies reveal that proinflammatory cytokines potentially contribute to OLs cell death. An overexpression of TNF-α was observed to induce OLs apoptosis which could contribute to the degenerative change and demyelination via TNFR1 and TNFR2 expressed in OLs (151). Other cytokines, such as IL-2, IL-1, IFNγ, and proNGF, all contribute to OLs apoptosis (142). In addition to apoptosis and necrosis, autophagy is activated in SCI, and Beclin1, a promoter of autophagy, is highly expressed in OLs (152).

Oligodendrocytes are fundamental to myelin formation as described earlier. The injury or death of OLs results in the degeneration of myelin sheaths and the support of axons by OLs would be disrupted after SCI which eventually lead to the widespread demyelination of spared axons. As a matter of fact, accumulating studies have demonstrated that demyelination indeed occurred in animal models and human after SCI (153, 154). For example, demyelinated axons were seen within 2 weeks after injury in paraplegic domestic animals in previous study (153). More interestingly, the extent of demyelination mainly contingents on the type and severity of injury. The normal myelinated axons are characterized by the regular distribution of sodium and potassium channels, after demyelination, the distribution of sodium and potassium channels is disrupted that contributes to an axonal conduction block (155). Besides, demyelination is identified to increase voltage-gated Na+ channels, which may result in Na⁺ influx during action potential propagation. To eliminate the excess Na⁺, more ATP is required which can disrupt the axonal internal energy balance. Additionally, the excess Na⁺ may lead to axonal Ca²⁺ overload via the Na⁺/Ca2⁺ exchangers. These events eventually result in axonal degeneration (156, 157). Besides, the demyelinated axons are vulnerable to damage in the microenvironment after SCI and ultimately lead to axonal degeneration (158).

After SCI, OPCs are multipotential stem cells which can differentiate into remyelinated cells to involve in axonal remyelination and contribute to the glial scar formation. McTigue et al. assessed the proliferation of NG²⁺ cells and OLs by bromodeoxyuridine incorporation and they found increased proliferation of NG²⁺ cells persisting throughout the first 4 weeks post-injury while the number of OLs continuously reduced by 7 days post-injury. However, they detected an increased number of OLs at 14 days post-injury. These results showed that proliferated NG²⁺ cells may differentiate into OLs after injury (159). Besides, the study using fate mapping confirmed that 30% of new OLs responsible for myelin regeneration were derived from OPCs while OPCs differentiate into the majority of myelinating Schwann cells (160). Other group also revealed that OPCs from the PDGFRα-expressing lineage could be transformed into functional myelinating Schwann cells after SCI (161). Moreover, by using genetic fate mapping, Hackett et al. found that ~25% of astrocytes were derived from NG²⁺ cells in the glial scar by 4 weeks after SCI (162). It is worth mentioning that the functions of OPCs are intricately modulated by a complex network including various factors (Table 4). We will not go into further discussion here. In addition to OPCs, the endogenous NPCs can also contribute to OLs replacement as they will be activated and migrate into the lesion after SCI (163, 164).

Remyelination occurs spontaneously on residual axons after SCI. Remyelination is difficult to detect until genetic fate mapping approaches are applied, the scientists can distinguish new myelin from preexisted myelin via labeling new myelination. Assinck et al. found that spontaneous remyelination was induced by OLs and myelinating Schwann cells in mice after SCI (160). Besides, other group detected remarkably clear visualization of spontaneously regenerated myelin in vivo (165). However, endogenous remyelination was limited due to multi-factors (166). Nashmi et al. found that the spontaneous remyelination in the injured white matter was non-optimal and incomplete because the newly formed myelin around the injured axons was thinner than normal myelinated axons (167). Recent studies have uncovered that multiple factors affected remyelination, such as (1) the myelinating OLs derived from OPCs are inadequate (168), (2) OLs maturation and myelination are limited (142), (3) axonal ensheathment and remyelination is influenced (169), and (4) OPCs, neural progenitor cells (NPCs) are affected by the unfriendly microenvironment (166). Therefore, more endogenous mechanisms of remyelination are needed to be explored.