Resident immune responses to spinal cord injury: role of astrocytes and microglia

Abstract

Spinal cord injury can be traumatic or non-traumatic in origin, with the latter rising in incidence and prevalence with the aging demographics of our society. Moreover, as the global population ages, individuals with co-existent degenerative spinal pathology comprise a growing number of traumatic spinal cord injury cases, especially involving the cervical spinal cord. This makes recovery and treatment approaches particularly challenging as age and comorbidities may limit regenerative capacity. For these reasons, it is critical to better understand the complex milieu of spinal cord injury lesion pathobiology and the ensuing inflammatory response. This review discusses microglia-specific purinergic and cytokine signaling pathways, as well as microglial modulation of synaptic stability and plasticity after injury. Further, we evaluate the role of astrocytes in neurotransmission and calcium signaling, as well as their border-forming response to neural lesions. Both the inflammatory and reparative roles of these cells have eluded our complete understanding and remain key therapeutic targets due to their extensive structural and functional roles in the nervous system. Recent advances have shed light on the roles of glia in neurotransmission and reparative injury responses that will change how interventions are directed. Understanding key processes and existing knowledge gaps will allow future research to effectively target these cells and harness their regenerative potential.

Introduction

Spinal cord injury (SCI) can be caused by a myriad of events and results in a debilitating loss of neurologic function. SCI occurs most frequently in the cervical, followed by the thoracic and thoracolumbar spine regions, and may be precipitated by any number of events, including automobile and sporting accidents, falls, and violence; moreover, SCI is becoming increasingly more common in the elderly population with pre-existing spine degeneration and spinal stenosis (Lenehan et al., 2012; Wu et al., 2013; Chen et al., 2016). Damage resulting from SCI may be diffuse and extend beyond motor and sensory dysfunction. Autonomic disruption following traumatic SCI may include gut dysbiosis, suppressed immune function, and endocrine disruption-establishing global injury that ultimately impairs neural repair (Ortega et al., 2023). Non-traumatic SCI, primarily occurring in the cervical and lumbar regions, is caused by one or a combination of myelopathic mechanisms such as arthritis, stenosis, or spondylosis that narrow the spinal canal and compress the cord. Symptoms differ based on the level of injury and may include neuropathic pain, numbness and tingling in limbs and extremities, partial or complete loss of sensorimotor function, and disruption of autonomic organ functioning such as bladder control. Overall, the incidence of traumatic SCI in North America is about 39 per million, and between 12–57 per million in Europe, with mortality rates between 10–20% within the first year (Ahuja et al., 2017; Barbiellini Amidei et al., 2022). Degenerative cervical myelopathy, specifically, is the most common form of spinal cord impairment worldwide, affecting 70% of individuals over 65 years of age (Badhiwala et al., 2020). While much of traumatic and degenerative SCI etiology is shared, degenerative SCI is characterized by its chronic progression, permitting vascular remodeling and plasticity as it develops (Blume et al., 2021). Comparatively, widespread axonal damage and extracellular matrix disruption occurring in the acute injury phase after traumatic SCI establish a permissive environment for neuroplastic changes (Sophie Su et al., 2016). This illustrates the multifaceted roles of glia in neuronal support and network rewiring, as well as lesion containment and inflammatory resolution.

While astrocytes (Sofroniew, 2015; Han et al., 2021) and microglia (Bellver-Landete et al., 2019; Brockie et al., 2021) have been extensively characterized and reviewed independently, this paper considers the two in parallel. This review synthesizes recent data on the critical roles played by astrocytes and microglia in wound corralling, lesion containment and regeneration (Sofroniew and Vinters, 2010; Zhou et al., 2020). We describe the function of microglia as first responders and phagocytes, in conjunction with astrocytes as regulators of vascular infiltration and immune cell recruitment.

Search Strategy

PubMed and Google Scholar were searched between January and March of 2022 with the following key words: Microglia, Spinal Cord Injury, Microglia and Inflammation, Microglia and ATP signaling, Purinergic Signaling, Microglial Phenotypes, Microglia and Synaptic Elimination, Gliotransmission, Glial Modulation, Astrocytes, Tripartite Synapse, Secondary Injury, Ischemia Reperfusion Injury. Papers were limited to publications in English, and abstracts were skimmed for relevance to key themes.

Primary and Secondary Injury Phases of Spinal Cord Injury

Spinal cord injury can be characterized by primary and secondary injury phases. Primary injury refers to the initial mechanical insult, causing cell death and disruption of the blood-brain/spinal-cord barrier (BBB/BSCB), resulting in instantaneous loss of function. This is most often the result of a contusion injury imposed from the posterior, resulting in a dorsal-to-ventral impact (Ahuja et al., 2017). Depending on the severity, injury to the dorsal grey matter results in sensory dysfunction including pain and numbness, while damage to the ventral grey matter results in functional motor impairment including loss of fine motor control and paralysis. Secondary injury occurs as vascular disruption causes immune cell infiltration and broad-scale inflammation. In traumatic SCI, this response is broadly initiated immediately following trauma, while the prolonged injury progression in degenerative SCI mitigates this process.

Treatment Approaches to Degenerative and Traumatic Spinal Cord Injury

Traumatic and degenerative SCI can be treated with surgical decompression to address cord compression. Despite its necessity in preventing further deterioration, surgical decompression may cause extensive ischemia-reperfusion (IR) injury in degenerative SCI, where large-scale blood flow is restored to compromised vasculature and tissues. This results in inflammatory responses as circulating leukocytes enter the injury site. This may be the cause of exacerbated disease that is observed post-operatively in 7–11% of surgical patients (Badhiwala et al., 2020). The prevention and treatment of IR injury is therefore an important component of degenerative SCI management. Evidence of IR injury following the restoration of blood flow has been characterized by bioenergetics disruption, free radical generation, necroptotic and apoptotic cell death, axonal degeneration, glial scarring, and cystic cavitation (Kalogeris et al., 2012; Sanderson et al., 2013; Vidal et al., 2017).

More broadly, depending on the level of injury, musculoskeletal as well as sympathetic neural functions may be lost. That is, injury at the medulla or cervical level may compromise vagal nerve control of cardiac function; injury at or above the thoracic level may impair sympathetic signaling to the heart and distal vasculature; and injury at the lumbosacral level may affect parasympathetic maintaining bladder and venous control (Ahuja et al., 2017). The remainder of this review will focus on the management and pathobiology of traumatic SCI.

Pharmacotherapeutic interventions

Current treatment for traumatic SCI, herein referred to as “SCI”, primarily centers around surgical decompression and acute hemodynamic management, followed by rehabilitative therapy. Early surgical decompression is recommended, with an emphasis on the acute time window (within 24 hours of injury). Within this phase, methylprednisolone sodium succinate transfusion may be offered within 8 hours of the injury, while anticoagulant treatment is recommended across the acute phase to attenuate the risk of thrombosis during hospital stay (Fehlings et al., 2017). Currently, there are no approved or widely adopted neuroprotective or neuroregenerative treatments for SCI. High-dose methylprednisolone has been the subject of extensive testing for its potential in preventing peroxidation of neuronal lipid membranes during acute secondary injury. Early clinical trials found an increased incidence of wound infection and lack of efficacy at both high (1000 mg) and standard (100 mg) doses. Subsequent trials failed to replicate negative effects and reported motor improvements up to 6 weeks post-operatively (Bracken et al., 1984, 1990, 1997), ultimately limiting the drug's clinical use to discretionary prescription by physicians.

Combinatorial pharmacotherapy may be used to modulate endogenous immune responses. Approaches could include: containing the spread of lesion sites through glial border formation (Sofroniew and Vinters, 2010; Gesteira et al., 2016), counteracting neuroinflammation, and attenuating secondary injury and excitotoxicity using, for example, intravenous immunoglobulin G and riluzole (Fehlings et al., 2012). The anti-inflammatory minocycline has also been used effectively in mice to counteract inflammation and enhance functional recovery compared to methylprednisolone (Wells et al., 2003), with moderate improvements recapitulated in cervical injury patients (Casha et al., 2012).

Regenerative approaches to neural injury

Regenerative approaches have largely focused on cellular transplantation to restore functional connectivity and remyelination. However, host glial cell responses to grafted cells are not well characterized.

Generation of neural progenitor cells for transplantation has been aided by recent approaches in direct reprogramming using transcription factors. This approach was first shown to convert fibroblasts into neurons and neural progenitor cells (NPCs) in vitro (Kim et al., 2011; Pang et al., 2011; Wapinski et al., 2013). Subsequent work has established the capacity of reprogrammed pluripotent stem cells, fibroblasts, and glial cells to create neurons, NPCs, and oligodendrocyte precursor cells (Guo et al., 2014; Pawlowski et al., 2017; Boshans et al., 2021; Kempf et al., 2021; Xu et al., 2023). Recent advances in small molecule-driven differentiation suggest alternative approaches to direct reprogramming that may promote more consistent, homogeneous cultures (Iyer et al., 2022). While this approach may better enable translation of direct reprogramming, time remains a limiting factor to its use, as small molecule reprogramming protocols take up to 3–4 times as long as transcription factor protocols.

Local delivery of trophic factors such as brain- and glial cell-derived neurotrophic factor (BDNF and GDNF, respectively) has demonstrated promising effects in pre-clinical models of SCI (Jain et al., 2006; Wang et al., 2008). NPCs have similarly been engineered to overexpress BDNF and GDNF (Blesch and Tuszynski, 2003; Khazaei et al., 2020; Li et al., 2023) to promote cell engraftment and axonal regrowth. Recently, NPC transplantation in conjunction with physical rehabilitation has enhanced functional gains after SCI (Lu et al., 2023). While these approaches show promise in rodent models, future work should focus on scaling up therapeutic approaches to large mammal and primate models to establish safety and efficacy.

Pathobiology of Spinal Cord Injury

At the initial insult, compression or transection of the spinal cord results in damage or death to neurons and oligodendrocytes, as well as activation of glial cells in the central nervous system (CNS) (Figure 1; Ahuja et al., 2017). Neutrophil infiltration peaks six hours post-injury and significantly declines in the remainder of the acute phase (Alizadeh et al., 2019). Nuclear factor-κB, tumor necrosis factor (TNF), and mitogen-activated protein kinase (MAPK) pathways are amongst those canonically upregulated post-injury, with significant dysregulation of MAPK microRNAs (Liu et al., 2020). Neurons suffer extensive damage both mechanically and metabolically. Axonal damage is exacerbated by damage to oligodendrocytes and disruptions in blood supply resulting in hemorrhage. This perpetuates inflammation, causing edema and broad-scale ischemia. In the ensuing secondary phase of injury, lasting up to 14 days, compromised blood flow creates hypoxic conditions that impair mitochondrial adenosine triphosphate (ATP) production and force neurons and other cell types to shift toward glycolytic ATP production, further impairing their ability to meet energetic demands (Kalogeris et al., 2012; Alizadeh et al., 2019). As this persists, cells release damage signals that recruit microglia and astrocytes to the site of injury. Here, astrocytes form a lesion border that contains debris and infiltrating leukocytes (Sofroniew and Vinters, 2010). Microglia, the resident immune cells of the CNS, aid in this process by corralling debris to the lesion core via Plexin-B2 signaling (Zhou et al., 2020). Further, microglia express and secrete both membrane-bound and soluble chemo- and cytokines that home to damage signals where they phagocytose cell debris and aid in containment of the injury core (Aloisi, 2001; Gaudet et al., 2011; Gülke et al., 2018; Bellver-Landete et al., 2019). As necrosis and apoptosis persist, cell contents including ATP and potassium are released into the extracellular matrix which, coupled with cytokine signaling, establishes damage-associated molecular patterns that recruit immune cells infiltrating the blood-spinal cord barrier (Brockie et al., 2021). Here, neutrophils, monocyte-derived macrophages, leukocytes, and B and T cells are allowed access to leaky vascular walls where their extravasation initiates the large-scale immune response that characterizes secondary injury and chronic inflammation (Anwar et al., 2016). The glial border is critical here to keep infiltrating cells to the lesion core and contain inflammation. In the following chronic phase of injury, resident immune cells play dynamic roles in mediating neuroinflammation, scar formation, and adaptive plasticity.

Figure 1.

Glial cells respond to injury by increasing the expression of transporters and trophic factors that serve roles in debris clearance, immune recruitment, border formation, and vascular permeability.

Adapted from Brockie et al. (2021) with permission. Created with BioRender.com. ACE: Angiotensin-converting enzyme; ANG-1: angiopoietin 1; ApoE: apolipoprotein E; AQP4: aquaporin 4; BBB: blood-brain barrier; CCL2: chemokine ligand 2; CD45: cluster of differentiation 45; CSPGs: chondroitin sulfate proteoglycans; CX3CR1: CX3C motif chemokine receptor 1; IBA-1: ionized calcium-binding adapter molecule 1; IL-1β: interleukin-1 beta; PGE2: prostaglandin E2; TGFβ: transforming growth factor-beta; TNF: tumor necrosis factor; TREM2: triggering receptor expressed on myeloid cells 2; Wnt: wingless-related integration site.

Microglial Function in Inflammation

Microglia originate in the CNS and function in tissue surveillance and injury responses. Following SCI, they initiate various response pathways to dynamically modulate the injured environment. Microglia regulate energy metabolism and debris clearance via purinergic signaling, recruit immune cells via antigen presentation, increase cytokine secretion, and monitor plasticity via synaptic elimination pathways.

Microglia, ATP, and purinergic signaling in neuroinflammation

Under physiologic conditions, microglia express both metabotropic (P2Y) and ionotropic (P2X) purinergic G protein-coupled receptors that bind nucleotides and maintain ATP balance (Calovi et al., 2019). Patch-clamping techniques show that microglia derived from rodent brains respond to extracellular ATP in culture (Walz et al., 1993) and contain ATP and adenosine-sensitive purinergic and P1 receptors (Langosch et al., 1994).

Following SCI, increased extracellular ATP serves as a damage-associated molecular pattern signal molecule as it is released from ruptured and apoptotic cells through a combination of exocytosis, leakage, and transporter-mediated release (Vénéreau et al., 2015; Calovi et al., 2019; Brockie et al., 2021). This in turn activates pro-inflammatory and phagocytic pathways (Elliott et al., 2009) via P2 signaling cascades (Langosch et al., 1994). Specifically, microglia express ionotropic P2X4, P2X7, and metabotropic P2Y6, P2Y12, and P2Y13 receptors (reviewed in Calovi et al., 2019). The presence of ATP-sensitive receptors on microglia allows early injury response activation. For example, Davalos et al. (2005) showed that after traumatic brain injury, ATP presence at the site of injury promotes the movement of microglial processes towards the injury site to form a protective barrier and that blockage of ATP signaling via an ATPase or P2Y receptor inhibitor significantly reduces their motility.

Microglial ATP signaling demonstrates dynamic effects on repair and debris clearance after neuroinflammation (Fiebich et al., 2014). Inhibiting ATP binding at P2Y receptors has been reported to reduce microglial outgrowth and impair cell elimination as well as phagocytosis (Davalos et al., 2005; Irino et al., 2008; Elliott et al., 2009; Ohsawa et al., 2010; Dissing-Olesen et al., 2014; Wendt et al., 2017; Madry et al., 2018). In addition to phagocytic modulation, ATP at sites of injury may be neuroprotective. N-methyl-D-aspartate-induced neuronal cell death can be mitigated if microglia are pre-treated with high concentrations of ATP, mediated by the P2X7 ionotropic receptor (Masuch et al., 2016). In the adult mouse brain, the presence of purinergic ligands (such as ATP) attenuated microglial activation in culture after lipopolysaccharide stimulation, evidenced by reduced expression of TNFα, interleukin (IL)-6, and IL-12 (Boucsein et al., 2003). This same protective effect has also been observed in the spinal cord, whereby ATP administration prevented the release of these cytokines following lipopolysaccharide stimulation (Ogata et al., 2003).

On the other hand, ATP can also induce neurotoxic responses after CNS injury. Lipopolysaccharide injection in the rat striatum has been found to upregulate the expression of P2X7 ionotropic receptors and increase several pro-inflammatory markers (Choi et al., 2007). Blockage of the P2X7 receptor using oxidized ATP as an antagonist subsequently reduced the number of apoptotic neurons and increased neuronal survival. The P2X7 receptor also plays a key role in interleukin release from cultured microglial cells, as ATP stimulation of the receptor increases mRNA expression of IL-6, TNFα, and chemokine CCL2 (Shieh et al., 2014). Given the role of microglia in ATP signal transduction, and of astrocytes in ATP generation and handling, the glial network may serve as a prominent target to further understand energy dynamics post-injury. This topic is reviewed in greater depth in Sections Astrocytes and Glial Modulation and Tripartite Synapse.

Microglial inflammatory and protective phenotypes

Similar to astrocytes, microglia have been classified into “classically activated” M1 versus “alternatively activated” M2 phenotypes based on key markers observed after injury (Kigerl et al., 2009, reviewed in more detail by David and Kroner, 2011). Briefly, it is believed that the M1 phenotype upregulates the expression of CD68, inducible nitric oxide synthase, and CD16/32 (David and Kroner, 2011), while the M2 phenotype expresses neuroprotective markers such as IL-10, arginase-1 and CD206 (Bedi et al., 2013). However, the M1/M2 notation does not define distinct phenotypes, but rather highlights two ends of the phenotypic spectrum following injury. Moreover, recent research in SCI using single-cell RNA sequencing suggests that microglia may adopt a much more general “disease-associated” phenotype after injury due to transcriptional reprogramming from a homeostatic state (Hakim et al., 2021), thereby refuting the idea that M1/M2 phenotypes are mutually exclusive. Rather, the recent investigation of microglia across development, aging, and injury has established between 9–12 clusters of transcriptionally distinct microglia that were recapitulated and localized in vivo by immunohistochemistry. Some of these include disease-associated microglia (Keren-Shaul et al., 2017), proliferative region-associated microglia (Li et al., 2019), white matter-associated microglia (Safaiyan et al., 2021), axon tract-associated microglia (Hammond et al., 2019), and others (reviewed in depth in Paolicelli et al., 2022).

Microglial recruitment and signaling

In addition to regulating debris and apoptotic clearance, microglia become the main antigen-presenting cells of the CNS following injury-induced activation (Carson et al., 1998). Microglia can present MHC Class II molecules and antigens to infiltrating CD4⁺ T-cells (Aloisi, 1999; Byram et al., 2004) while also stimulating the proliferation of Th1 (INF-y producing) and Th2 (IL-4 producing) CD4⁺ T-cells (Chastain et al., 2011).

Furthermore, microglia can upregulate signaling molecules like the fractalkine receptor, CX3CR1, to increase neuronal interactions following injury. In the CNS, CX3CR1 is predominantly expressed by microglia as a surface receptor, while the ligand, CX3CL1 or “fractalkine”, is primarily expressed by neurons (Harrison et al., 1998; Wolf et al., 2013). The fractalkine pathway, also known as “neurotectin”, is part of the CX3C chemokine family and serves as a homing signal for phagocytes. CX3CL1 can exist in membrane-bound form, expressed by neurons, or in soluble form, both of which recruit microglia expressing the receptor, CX3CR1 (Sokolowski et al., 2014). In the quiescent state, this ligand exists in its membrane-bound isoform, while injury causes its upregulation by neurons, followed by cleavage by a disintegrin and metalloproteases (ADAM; Hundhausen et al., 2003; Li et al., 2015; Wang et al., 2017). Here it exists as a soluble cytokine that serves as a homing signal for microglia and macrophages. Although fractalkine is needed for microglial recruitment, it appears that in the context of SCI, CX3CR1 deficiency may be beneficial for functional recovery and synaptic plasticity. Donnelly and colleagues (2011) demonstrated improved functional recovery in CX3CR1 knockout animals following SCI, paralleled by reduced recruitment of monocyte-derived macrophages that typically contribute to the release of cytokines and oxidative species. Absence of CX3CR1 has also been found to enhance motor neuron plasticity in the lumbar ventral horn and reduce pain responses in the setting of spinal and sciatic nerve lesions (Zhuang et al., 2007; Staniland et al., 2010; Freria et al., 2017). Conversely, in models of multiple sclerosis, infusion of soluble fractalkine was found to attenuate microglial/macrophage activation and enhance remyelination by oligodendrocytes and oligodendroglial precursor cells (de Almeida et al., 2023), thus illustrating the potential of the fractalkine pathway as a dynamic therapeutic target in neural injury.

Microglia and synaptic elimination

Microglia constantly survey their environment by extending processes that make contact with synapses (Davalos et al., 2005; Wu et al., 2007; Wake et al., 2009; Kettenmann et al., 2013). This surveying behavior allows microglia to regulate synaptic elimination, which directly impacts functional recovery after SCI. During development, the engulfment of synapses by microglia, or “pruning”, is necessary for normal brain development and sensory function (Schafer et al., 2013; Paolicelli and Ferretti, 2017; Gunner et al., 2019 and reviewed in depth by Andoh and Koyama, 2021). Interestingly, the opposite appears to be true following SCI. In contrast to development where pruning is critical for normal function, it has been observed that pruning (referred to as “elimination” in adulthood) of synapses is maladaptive following SCI. Takano and colleagues (2014) observed that twy/twy mice, which develop severe compression of the spinal cord over time, exhibit an upregulation of C1q: a member of the classic complement-dependent synaptic elimination pathway. Using immune-electron microscopy, they concluded that the neurodegeneration observed after SCI is due to microglia making direct contacts with pre- and post-synaptic structures to initiate engulfment. In traumatic SCI, Freria et al. (2017) demonstrated that the absence of CX3CR1 (a key component of the chemokine-based synaptic elimination pathway) increased the formation of immature dendritic spines and excitatory/inhibitory synapses in mice, reflecting a greater ability to endogenously repair damage after injury. These findings suggest maladaptive effects of synaptic elimination after SCI. This idea is further supported by the fact that breathing circuitry can be endogenously repaired after non-traumatic SCI via synaptic connections established by mid-cervical excitatory interneurons (Satkunendrarajah et al., 2018). One might therefore propose that inhibiting synaptic elimination may allow this same bridging and reparative mechanism to occur in motor circuits to preserve function following spinal injury.

Microglia and Pain Modulation

Nerve injury, including SCI, peripheral nerve injury, and even diabetic neuropathy, elicits pro-inflammatory signaling that is intricately linked to neuropathic pain and mechanical allodynia. Multiple changes in expression occur in activated microglia that affect neuronal signaling. Monocyte chemoattractant protein-1 (Tanaka et al., 2004), involved in the recruitment of blood-derived macrophages, and matrix metalloproteinase-9 (Shubayev et al., 2006), affecting extracellular matrix integrity, have been implicated in microglial activation following peripheral nerve injury. Mice deficient in monocyte chemoattractant protein-1 and matrix metalloproteinase-9, respectively, have shown reduced microglial activation and subsequently attenuated allodynia. Conversely, intrathecal injection of these proteins enhances activation and allodynia, potentially mediated by upstream signaling effects of fractalkine (CX3CL1), IL-1B, and TNFα (Shubayev et al., 2006; Tsuda, 2016).

Additional changes contributing to the microglial disease phenotype involve the expression of purinergic receptors, MAPKs, and cytokines as discussed previously. Purinergic signaling involving porous P2X and G-protein coupled P2Y receptors drives ATP transmission and signaling to afferent neurons via ionic gradients. P2X4R, specifically, stimulates the microglial release of BDNF, which then alters anionic gradients across dorsal horn motor neuron membranes, creating a depolarizing effect causing hyper-responsivity (Ulmann et al., 2008; Trang et al., 2009; Beggs et al., 2012). P2X4R is upregulated in microglia after peripheral nerve injury and accordingly, its inhibition mitigates allodynia, which intrathecal administration therein increases (Tsuda et al., 2013). Upstream of P2X4R, interferon regulatory factor-5 and interferon regulatory factor-8 are transcription factors modulating P2X4R upregulation and their inhibition correspondingly mitigates allodynia by this pathway (Masuda et al., 2012; Tsuda et al., 2013).

MAPKs p38 and extracellular signal-regulated protein kinase (ERK) expression are altered following nerve injury. p38 is upregulated and activated by the purinergic receptor P2Y12R (Kobayashi et al., 2008). Similarly, ERK expression is upregulated immediately after nerve injury, mediated first by dorsal horn neurons, followed by microglia and astrocytes in the days and weeks following (Zhuang et al., 2005). Similar to monocyte chemoattractant protein-1 and matrix metalloproteinase-9, IL-1B, TNFα, and IL-6 contribute to the activation of ERK and inhibition of both ERK and p38 attenuate neuronal hyper-excitation and chronic pain (Tsuda, 2016). Additionally, inhibiting p38 reduces IL-1B expression (Ji and Suter, 2007). Taken together, these patterns illustrate the dynamic feedback regulation of microglial inflammatory signaling factors that subsequently drive allodynia and chronic pain cascades. While this entwinement makes it difficult to determine the mutual exclusivity of the two processes, it highlights the value of microglia as a therapeutic target in SCI and pain modulation.

Broadly, microglial activation may drive the experience of pain in one of two possible ways. Firstly, it may exacerbate pain by increasing inflammatory signaling, causing secondary injury that damages adjacent neurons and axons that are the basis for neuropathic pain. Alternatively, activated microglia upregulate synaptic pruning, resulting in the elimination of neuronal synapses that are substrates for painful sensory memories (Ward and West, 2020). Degrading these synapses may consequently prevent their persistent reactivation and protect against chronic pain. Neutrophil activation has, however, been shown to be protective against the transition from acute to chronic lower back pain in humans (Parisien et al., 2022), suggesting a critical role for glial activation and neuroinflammation. Due to the complex and rapidly changing inflammatory milieu that follows nerve injury and SCI, it is difficult to evaluate the effects of individual signaling pathways independent of the various cell-mediated responses occurring around them. Given the role of microglia in signaling to dorsal horn afferents, and their close integration with other neural and circulatory cell types like astrocytes and leukocytes, future research may interrogate this population to understand key biological effectors of chronic pain and potential therapeutic targets.

Astrocytic Function in Inflammation

In injurious conditions, astrocytes serve critical roles in fluid and ion balance, blood flow, and the formation of anatomical barriers in both development and injury. Astrocytic processes are densely packed with water channels, namely aquaporin 4, and multiple ion channels including Na⁺/H⁺ and K⁺ transporters that work in conjunction with aquaporin 4 to maintain fluid and ion homeostasis. Astrocytic processes also extend endfeet to the BSCB, where they maintain endothelial integrity and regulate blood flow in neural tissue. Prostaglandins, arachidonic acid, and nitric oxide are all produced by astrocytes to affect changes in vessel diameter and flow in response to changes in activity and demands, as well as deviations from homeostatic conditions (Gordon et al., 2011). In the subacute period, reactive astrocytes serve critical lesion containment roles. By 2–3 weeks post-injury, a compact astrocytic-microglial scar composed of chondroitin sulfate proteoglycans (CSPGs) has formed around the lesion core in which leukocytes, macrophages, and astrocytes are contained, while spared neurons and oligodendrocyte precursor cells may be found in the perilesional tissue (Wanner et al., 2013; Ren et al., 2017). The compromised BSCB is resealed by astrocytes shortly thereafter (Bush et al., 1999). In the following weeks, extracellular matrix remodeling and regenerative processes like remyelination and axonal regrowth may take place depending on the extent of spared tissue. Regenerative potential is significantly limited by aging and these processes thus represent key areas for therapeutic intervention to optimize functional recovery.

Astrocytes Regulate Blood-Brain/Spinal Cord Barrier Permeability

Astrocytes play major roles in injury as they are chief regulators of inflammation through their maintenance of the BBB and the containment of lesion sites (Sofroniew, 2015). These functions are necessary to maintain the relative immune privilege of the CNS that limits inflammation and protects post-mitotic neural cells from extensive secondary damage (Harris et al., 2014). Astrocytes, while only reported to be involved postnatally, play critical roles in the maturation and maintenance of the BBB. Though the mechanisms of permeability regulation are incompletely understood, fundamentally, astrocytes regulate permeability by modulating endothelial tight junctions via metabotropic signaling (Obermeier et al., 2013). Major effectors include apolipoprotein-E (ApoE), Wnt signaling, angiotensinogen, and Sonic Hedgehog (Milsted et al., 1990; Alvarez et al., 2011).

Allelic variation in ApoE affects the protein form secreted by astrocytes and has been implicated in various neurodegenerative diseases affecting both the brain (e.g. Ballard et al., 2004; Belloy et al., 2019) and spine (Desimone et al., 2021). ApoE4, specifically, has been shown to compromise astrocyte end-feet contact with vasculature, resulting in leaky vessels (Jackson et al., 2022). Astrocytic Wnt signaling is also involved in mediating end-feet contacts, such that loss of astrocytic Wnt secretion exacerbates age-related decline (Guérit et al., 2021).

Angiotensinogen is converted into its active form, angiotensin-I (ANG-I), by renin before being converted by angiotensin-converting enzyme into ANG-II, which drives tightening of vascular barriers (Wosik et al., 2007). What mediates astrocytic expression of these effectors remains incompletely understood but soluble factors including IL-6, GDNF, and fibroblast growth factor 2 have been reported to participate in the induction of BBB integrity (Sobue et al., 1999).

Vascular permeability in homeostasis and injury

Under homeostatic conditions, astrocyte-released ANG-II maintains BBB integrity by promoting the shuttling of tight junction proteins JAM-1 and occludin to lipid rafts (Wosik et al., 2007). In contrast, injury results in the downregulation of ANG-II via various cytokine signaling mechanisms, compromising BBB integrity (Wosik et al., 2007; Alam et al., 2020). Activated T cells are among the first circulating cells to cross the BSCB, with numbers peaking by 9 hours post-injury and returning to baseline over the next two days (Hickey et al., 1991). Circulating neutrophils are then attracted to lesion sites where their membrane-bound integrins bind endothelial intracellular adhesion molecules 1 and 2, allowing them to contact with the BBB within 12 hours of injury (Carlos et al., 1997; Kim et al., 2009). Here, they release metalloproteases, TNF, and reactive oxygen species that further dissolve endothelial integrity, ultimately facilitating their extravasation along with monocytes and other immune cells (Scholz et al., 2007). Monocytes have been shown to then recruit astrocytes via chemokine signaling, specifically CCL-2, where they then extend processes toward these cells at the lesion site, aiding in its containment and glial scar formation (Choi et al., 2020). Later in the recovery phase, Sonic hedgehog secretion is upregulated in astrocytes, where it can then bind to endothelial cell-bound receptors to reinstate BBB integrity (Alvarez et al., 2011). The roles of astrocytes in immune recruitment, border formation, and vasculature integrity make them critical effectors across the acute to chronic time course and illustrate the need to understand their activation phenotypes and response mechanisms.

Astrocytic Inflammatory and Protective Phenotypes

Gross-scale consideration of astrocyte roles in inflammation in the past has led to the ascription of polarized phenotypes, “A1” and “A2”, defined by either a reactive, pro-inflammatory, neurotoxic profile or an anti-inflammatory, neuroprotective one. While these subtypes were not intended to create a binary representation of astrocytic phenotype, their misinterpretation led to much debate and consequential repudiation of their validity. Similar to the discourse following the M1/M2 macrophage polarization paradigm, the A1/A2 descriptions are criticized for their narrow, absolutist definitions. That is, not only does this binary model not encompass the full range of cellular phenotypes, but it describes the two as being mutually exclusive when in fact, canonical pro- and anti-inflammatory markers are often co-expressed within singular astrocytes (Escartin et al., 2021). It may be more accurate, therefore, to refer to astrocytic phenotypes as poles of a continuous spectrum.

Astrocytic markers in various environments

Following spinal cord injury, astrocyte reactivity is induced by microglial cytokine signaling, primarily via IL-1α, TNF, and complement component 1 subcomponent q (Liddelow et al., 2017). These astrocytes were observed to have subsequently decreased phagocytic activity and dramatically upregulated complement component 3 expression, likely in correlation with impaired synaptic formation (Liddelow et al., 2017). Other canonical protein markers of reactivity include Stat3 (Okada et al., 2006), glial fibrillary acidic protein, and vimentin (Liu et al., 2014). In contrast, quiescent astrocytes have been reported to upregulate neurotrophic factors promoting cell growth, as well as form glial lesion borders and restrict the spread of inflammation (Sofroniew and Vinters, 2010; Liddelow et al., 2017).

Transcriptional accessibility of genes has also been measured to infer response phenotypes, but these do not directly correlate to differentially expressed genes in neuroinflammatory and SCI models (Burda et al., 2022). Moreover, changes in protein expression of markers like glial fibrillary acidic protein can also result from exercise and environmental changes (Rodríguez et al., 2013). These discrepancies highlight the need to correlate changes across multiple target markers. Further, it is critical that phenotypic markers are examined across multiple modes of expression in vivo and in situ, as enzymatic digestion of tissues used to isolate cells for characterization may induce changes in expression (Marsh et al., 2022).

Astrocytes and Glial Modulation

The nature and magnitude of the role of astrocytes in the tripartite synapse have been questioned in the context of physiologic and inflammatory conditions. Gliotransmission posits that astrocytes release neurotransmitters via Ca²⁺-dependent signaling that act directly on neurons. Originally, these transmitters, termed “gliotransmitters” were thought to include D-serine, ATP, and glutamate (Halassa et al., 2007). D-serine has since been contested as an astrocyte-derived information conduit, as its release and uptake, when studied in greater depth, were found not to coincide directly with propagating calcium waves (Wolosker et al., 2016). Further, astrocytes themselves appear to synthesize L-serine, not D-serine; indeed, astrocytes have not been found to produce serine racemase, the enantiomer-converting enzyme needed to yield D-serine. Instead, neurons are known to produce serine racemase, suggesting their role in the final step of D-serine synthesis (Wolosker et al., 2016). This model, which depicts L-serine being produced by astrocytes, shuttled to neurons, and then converted into D-serine, provides an illustration of the prominent exchange of serine between astrocytes and neurons and supports the role of neurons as primary modulators of plasticity in learning and development.

The roles of ATP and glutamate as neuronal signalers, however, remain less understood and leave the validity of gliotransmission in question. For instance, it has been shown that astrocytes in vitro respond to synaptic neurotransmission with wave-like calcium transmission across the cell and synapses and that these calcium transients may subsequently evoke their exocytotic release of ATP (Koizumi, 2010). Variations in [Ca²⁺] in astrocytes have also been reported to elicit their glutamate release, specifically in response to stimulation of G-protein coupled metabotropic glutamate receptors (Bezzi et al., 2004). The extent to which calcium flux equates to gliotransmissible signals that subsequently induce a neuronal response, however, has been questioned with regard to discrepancies in the timing and synchronization of transients between astrocytes and neurons (Fiacco and McCarthy, 2018). Recent advances in microscopy may facilitate the measurement of glial transmitter release across the tripartite synapse using cell-specific stimulated emission depletion imaging (Heller et al., 2020; Arizono and Nägerl, 2022).

Tripartite Synapse

In both injurious and homeostatic conditions, astrocytes are intricately connected with neuronal function in the context of the tripartite synapse, where they modulate neurotransmission between pre- and postsynaptic neurons (Figure 2). Injury-induced breakdown of the BSCB facilitates the extravasation of glucose, resulting in increased extracellular glutamate concentrations. With excessive and prolonged synaptic glutamate binding, the influx of calcium results in the production of nitric oxide and reactive nitric and oxidative species (Greene and Greenamyre, 1996). As ischemic conditions spread with swelling, reactive nitric and oxidative species and ionic membrane imbalance further impair mitochondrial function in a stepwise fashion, such that mitochondrial ATP production is increasingly dampened 2, 6, 12, and 24 hours after injury, indicating persistent damage leading to secondary injury (Jia et al., 2016). As a result, oxidative phosphorylation is inhibited and cytosolic glycolysis begins to take over, further contributing to energy depletion and cell dysfunction (Orrenius, 2007; Sanderson et al., 2013).

Figure 2.

Astrocytes are in close contact with neurons at the tripartite synapse where they support their function through glutamate cycling, secreting trophic factors (e.g.: neuroligins, neuregulins, etc.), and metabolic support.

Created with BioRender.com. ATP: Adenosine triphosphate; GLN: glutamine; GLT: glutamate; GS: glutamine synthetase; EAAT1/2: excitatory amino acid transporter 1/2.

Astrocytes aid in this process by providing metabolic support via multiple pathways. Firstly, astrocytes can efficiently switch to glycolysis in hypoxic conditions to preferentially catabolize pyruvate to lactate via dehydrogenase 5, which can then be used by neurons in energy production (Bélanger et al., 2011). While many ATP-generating cell types are separated from the blood supply by three cell layers, astrocytes are in direct contact with neural vasculature, affording them direct access to glucose and oxygen (Mathiisen et al., 2010). They also operate with relatively low metabolic demands (Magistretti et al., 1999), allowing them to efficiently generate ATP that can be used to supplement neuronal demands (Beard et al., 2022). This ATP has also been found to act on neuronal A1 receptors to inhibit adenylyl cyclase-mediated neuronal apoptosis (Pellerin and Magistretti, 1994; Boison, 2006; Brooks, 2018). Conversely, aberrantly released ATP from damaged cells may serve as a molecular alarmin to microglia, which in turn elicits a pro-inflammatory response (Gülke et al., 2018).

In addition to supplementing ATP production, astrocytes support neuron signaling by recycling neurotransmitters. Given that 80–90% of synapses are excitatory (Danbolt, 2001), this represents an abundance of glutamate. Astrocytes and neurons utilize synaptically-released glutamate for several functions including metabolic oxidation and transmitter synthesis (McKenna, 2007; Ayers-Ringler et al., 2016). The glutamate-glutamine system refers to the synthesis of glutamate and gamma-aminobutyric acid precursors from metabolically compatible glutamine. Astrocytes and neurons utilize synaptic glutamate for several purposes and its metabolic usage may be compartmentalized across distinct brain regions (Carmignoto and Fellin, 2006; McKenna, 2007; Platel et al., 2010). In astrocytes, glutamine synthetase, pyruvate carboxylase, cytosolic malic enzyme, and mitochondrial branched-chain aminotransferase enzymes all participate in glutamate metabolism, while glutamate aspartate transporter/excitatory amino acid transporter 1 (EAAT1) and glutamate transporter 1/EAAT2 are responsible for glutamate transport into and out of these cells. Conversely, in neurons, pyruvate dehydrogenase, phosphate-activated glutaminase, mitochondrial malic enzyme, and cytosolic branched-chain aminotransferase facilitate metabolism and EAAC1/EAAT3, EAAT4, and EAAT5 glutamate transporters regulate flux. In both neurons and astrocytes, aspartate aminotransferase and glutamate dehydrogenase also aid in the metabolic process (McKenna, 2007; Pajarillo et al., 2019). In physiologic conditions, glutamate transporters are responsible for glutamate reuptake during neurotransmission when concentrations increase rapidly from 100 μM to 1 mM (Danbolt, 2001; Mahmoud et al., 2019). In injurious conditions and inflammation, excitotoxicity occurs when excess glutamate released from glia and neurons accumulates as a result of transport dysfunction and lack of feedback control, thereby disrupting ionic balance.

Conclusion

In summary, spinal cord injury imposes primary, mechanical injury, followed by secondary injury and extensive inflammation. Resident immune cells are first responders to acute injury and chief regulators of immune responses in the chronic phase. While glia are responsible for inflammatory signaling that often imposes secondary injury, this response cannot be dismissed as detrimental and one to abolish. Astrocytes mediate vascular permeability, form lesion borders, and maintain energy dynamics across neurons. Microglia corral and phagocytose cell debris as well as recruit circulating macrophages and leukocytes. In conjunction with astrocytes, they are critical to lesion containment. Following injury, the scarring formed across the lesion, formerly termed the ‘glial scar’, offers structural support for potential axon regrowth. The absence or dysfunction of any of these elements results in diffuse inflammation and inhibits regeneration.

Traumatic injury causes mechanical damage to these networks, compromising their structural integrity and creating a harsh inflammatory environment that causes prolonged cellular degradation. To optimize functional outcomes, we must untangle the region-specific responses of glial cells across time. Continued efforts to characterize microglial and astrocyte profiles using deep single-cell RNA sequencing and in vivo approaches will thereby identify novel therapeutic targets in SCI treatment, as well as offer translational knowledge to various neural pathologies. Such advances enhance treatment capabilities and will ultimately improve patient outcomes.

In response to major paradigmatic shifts in the glial biology field over the past decade, this review defines the roles of astrocytes and microglia as both independent and integrated cell networks in the context of SCI and inflammation. By summarizing key areas of debate including M1/M2 and A1/A2 polarization, and gliotransmission, we highlight current knowledge and outstanding questions in glial biology.

Additional file: Open peer review report 1 ^{(113.1KB, pdf)} .

Funding Statement

Funding: This work is supported by the Robert Campeau Family Foundation/Dr. C.H. Tator Chair in Brain and Spinal Cord Research (to MGF).