The Cortical NG2-glia Response to Traumatic Brain Injury

Abstract

Traumatic brain injury (TBI) is a significant worldwide cause of morbidity and mortality. A chronic neurologic disease bearing the moniker of “the silent epidemic”, TBI currently has no targeted therapies to ameliorate cellular loss or enhance functional recovery. Compared to those of astrocytes, microglia, and peripheral immune cells, the functions and mechanisms of NG2-glia following TBI are far less understood, despite NG2-glia comprising the largest population of regenerative cells in the mature cortex. Here, we synthesize the results from multiple rodent models of TBI, with a focus on cortical NG2-glia proliferation and lineage potential, and propose future avenues for glia researchers to address this unique cell type in TBI. As the molecular mechanisms that regulate NG2-glia regenerative potential are uncovered, we posit that future therapeutic strategies may exploit cortical NG2-glia to augment local cellular recovery following TBI.

1. INTRODUCTION

Having earned the moniker of “the silent epidemic”, traumatic brain injury (TBI) afflicts an estimated 69 million individuals worldwide each year¹. This has significant global impact. For instance, in the United States, 2.8 million people suffer TBI yearly, including 630,000 children^2,3. It is the leading cause of death in 1-44 years olds^2–4, and its economic cost totals $102bn yearly at current day value². As TBI causes persistent physical, cognitive, and psychological disabilities in survivors, it is appropriately considered a chronic neurologic disease; in fact, TBI and Alzheimer’s Disease (AD) may be pathophysiologically linked^5,6. However, no targeted therapies exist for TBI: all patients, young and old, can only be provided with supportive care and rehabilitation. Therefore, there is a critical need to uncover post-injury cellular mechanisms that may have therapeutic targets to aid in recovery from TBI⁷.

Compared to the post-TBI cellular response of resident glial cells (e.g. microglia, astrocytes) and members of the innate (e.g. neutrophils, monocytes) and adaptive immune systems (e.g. T-cells)^8–10, the response of NG2-glia, also commonly referred to as oligodendrocyte precursor cells (OPCs), is far less researched. At 4-8% of the total number of cells, NG2-glia comprise the largest population of regenerative cells in the adult CNS¹¹, accounting for 70-80% of active cellular proliferation in a naïve, uninjured brain^12,13. NG2-glia also likely serve in cellular regeneration following TBI, demonstrating response kinetics for polarization, migration, and proliferation within 24 hours after an injury and persisting for at least one month^12,14,15, therefore bridging the multiple phases following an injury: the “acute” or “early” phase (e.g. 0-2 days after injury), the “subacute” or “intermediate” phase (e.g. days to weeks after injury), and the “chronic”, “late”, or “delayed” phase (e.g. months to years after injury)¹⁶.

NG2-glia in the mature brain are perhaps most well known as a contributor to repair mechanisms in demyelinating diseases (e.g. multiple sclerosis^17–19) and other white matter injuries (e.g. neonatal hypoxic ischemic encephalopathy²⁰). This is justified by the proclivity of white matter NG2-glia to differentiate into oligodendrocytes, which contribute to remyelination²¹. Similarly, as TBI induces diffuse white matter injury with significant clinical implications^22,23, there is interest in investigating the role of NG2-glia in post-TBI remyelination as well^24,25. However, TBI also encompasses a wide range of focal cortical pathologies (e.g. skull fracture, parenchymal laceration, contusion, hemorrhage, focal edema)²⁶ which have been associated with early disease progression and long-term neurologic outcomes^27–30. Because NG2-glia demonstrate motility, proliferative capacity, and potential for multipotency soon after injury¹¹, they present a unique lens by which to evaluate the cellular mechanisms contributing to the evolution of cortical trauma. Additionally, as conventional imaging techniques employed in the diagnosis of acute TBI (i.e. computed tomography) are more likely to identify focal cortical pathologies, especially if accompanied by hemorrhagic features, than other diffuse or white matter injuries, there is a seductive possibility that early exploitation of cortical NG2-glia endogenous regenerative could serve as a therapeutic strategy in the future. It is our goal here to review the relevant literature on the post-TBI responses of cortical NG2-glia, specifically, and the known molecular mechanisms that underly these cellular functions.

2. CORTICAL NG2-GLIA: A FUNCTIONALLY DISTINCT POPULATION

A relatively recent discovery, NG2-glia were initially identified as non-neuronal cells lacking the usual structural components of the three other major glial classes: astrocytes, oligodendrocytes (OLs), and microglia³¹. Today, this population is most often characterized by the co-expression of OL markers (e.g. Olig2, Sox10)^13,32 in addition to two proteins not expressed by mature OLs: platelet derived growth factor receptor alpha (PDGFRA) and neuron-glial antigen 2 (NG2, also called chondroitin sulfate proteoglycan 4 [CSPG4])³³. Expression of PDGFRA precedes that of NG2 in NG2-glia, while expression of both markers decreases as these cells differentiate into mature OLs^33,34. Of note, NG2, an integral membrane chondroitin sulfate proteoglycan^35,36, is expressed by CNS cell types, such as pericytes, microglia, and macrophages, especially in the setting of pathology^37–40, while PDGFRA is also expressed by certain vascular and leptomeningeal cells⁴¹.

NG2-glia are found at seemingly comparable densities throughout the brain, including cortical gray matter, white matter, and deeper structures⁴². As they can self-renew and are multipotent, they also populate neurogenic areas such as the subventricular zone (SVZ) and dentate gyrus. Although NG2-glia are widespread, it has become increasingly apparent that they do not comprise a homogenous population, rather, demonstrate different properties and functions dependent on spatial location within the brain and the developmental stage of the organism¹¹. For instance, the cell fates of NG2-glia isolated from white matter differ from those isolated from gray matter, regardless if they are heterotopically transplanted to gray matter and white matter, respectively⁴³. While an initial study could not differentiate distinct subpopulations of NG2-glia between gray and white matter⁴¹, a more recent study has uncovered different proportions of NG2-glia subpopulations both at rest and following injuries, including the upregulation of specific genes in cortical NG2-glia versus their gray matter counterparts (Wnt signaling pathway: Fzd8, Axin2, AMPA receptor subunit: Gria2)⁴⁴. Therefore, findings from injury models involving white matter NG2-glia^{15,17,19,45–47} may not be generalizable to cortical NG2-glia and vice versa. Similarly, progenitors from the SVZ are likely another unique population of cells that can migrate towards an injury site and contribute to the perilesional cellular response. However, the kinetics of their arrival to the injury site are thought to be slower than the immediate responses of perilesional cortical NG2-glia and are complicated by whether the SVZ and rostral migratory streams are also affected by the injury, themselves^48–50. For the purpose of this review, we will focus primarily on the role of cortical NG2-glia in the response to TBI.

There are a number of functional characteristics that make NG2-glia a unique population of cells in the mature cortex. The first is that they are capable of ongoing proliferation and differentiation into other cell types. However, unlike their white matter counterparts, of which 80-90% will eventually differentiate into mature OLs, cortical NG2-glia most often continue to generate more NG2-glia, both in naïve and injured brains^51,52. They also do not readily differentiate into astrocytes as seen during development^53–55. In the mature, naïve animal, the majority of cortical NG2-glia (>96%) reside in a prolonged G1 phase, with only a miniscule percentage demonstrating proliferation at a rate that closely balances the rate of NG2-glia differentiation (e.g. into mature oligodendrocytes) and apoptosis, permitting a constant NG2-glia cell density over time^14,52. Conversely, in the setting of injury or other environmental stimuli (e.g. exercise), cortical NG2-glia can respond quickly, either by re-entering the cell cycle or exiting the cell cycle and differentiating, respectively^14,52.

A second noteworthy feature of cortical NG2-glia is that these cells are structurally dynamic. At a baseline state, they have a stellate configuration, with processes that extend/retract over the course of minutes¹⁴. They are also motile and migrate approximately 2 mm per day¹⁴. This motility is essential for maintaining an even density throughout the naïve cortex, as NG2-glia appear to divide and migrate towards areas in which other NG2-glia may have undergone apoptosis or differentiation¹⁴. Furthermore, upon acute application of a destructive stimulus (e.g. laser ablation, stab wound), NG2-glia demonstrate acute hypertrophy and polarization with process extension and migration towards the injury on a time course of several hours^14,56. The NG2-glia responses following injury is perhaps most similar to the those of microglia, which activate, hyperpolarize, and migrate towards the injury on the order of minutes, and proliferate within 24 hours^10,57. Conversely, astrocytes, which demonstrate significant astrogliosis and have functions in regulating blood-brain-barrier integrity and immune signaling following brain injury, demonstrate minimal gross migration following injury^10,58.

A third unique NG2-glia characteristic, and one that is perhaps the least explored, is that these cells appear to play a role in cell-to-cell signaling, as they express a variety of voltage-dependent and ligand-dependent ion channels and receive excitatory and inhibitory inputs from neurons^59–61. These cortical neuron-NG2-glial synapses are critical in regulating NG2-glia functions and differentiation^62–65, however their role in the adult cortex and in response to TBI is unexplored.

3. ANIMAL MODELS OF TBI

TBI is not a single disease entity, rather, an umbrella diagnosis that encompasses multiple types of brain pathology following head trauma. Within TBI are focal (e.g. contusions, intraparenchymal hemorrhage) and diffuse (e.g. traumatic/diffuse axonal injury, hypoxic ischemic injury, microvascular injury) pathologies^26,66,67, which can present singly or in combination. These pathologies are most dramatic and often found together in severe TBI²⁶, but even mild TBI is characterized by focal, injury-adjacent^68,69 and deep white matter^70,71 pathologies. Multiple animal models have been developed to study TBI, each recapitulating a unique combination of the injury characteristics seen in human neurotrauma^72–75, including stab wound injury, open head injury, closed head injury, and rotational injury models (Figure 1). However, as cortical NG2-glia have been studied in various rodent TBI models with different pathological features (Supplemental Table 1), findings from one experimental setup may neither be applicable to other models nor generalizable to human TBI.

Figure 1. Models of Rodent TBI.

(A) In the stab wound injury (SWI) model, a sharp blade or needle is precisely inserted through a surgically created craniotomy into cortex (or cerebellum) and withdrawn. While the injury is dependent on the size and depth of the blade, this is most often a focal, mild TBI.

(B) In “open head” injury models (e.g. controlled cortical impact, weight drop, fluid percussion injury), a mechanical force is directly applied to the brain parenchyma through a surgically created craniotomy.

(C) In “closed head” injury models, the mechanical force is applied to the skull, either with or without a scalp incision. While more closely resembling clinical TBI, this model more often provides more heterogenous injuries than the open model (B).

(D) Rotational acceleration injury models do not direct force towards the brain/skull, rather, rotate the brain relative to the skull around a fixed axis. Some models require a mount affixed directly to the skull, while others are capable of rotation of a freely mobile head. The result is tissue shear and axonal injury, with minimal direct contusional pathology.

Created with BioRender.com

In the cortical stab wound model of TBI, the cranial vault is surgically breached (e.g. burr hole, craniectomy), and a sharp blade or needle is precisely inserted into cortex (or cerebellum) and withdrawn. This produces neuronal and vascular injuries that most closely resemble the direct tissue injury of a neurosurgical procedure; this is in contrast to the higher energy direct, indirect, and rotational forces of other penetrating, closed head, or rotational TBI models. This model results in a highly reproducible focal cortical injury that is most similar to “mild” injury induced by controlled cortical impact models discussed below⁷⁶.

“Open head” TBI describes animal models in which a high energy force directly injures the brain through a surgical craniotomy, with or without replacement of the skull flap. Prime among these is the controlled cortical impact (CCI) model, which utilizes a mechanical piston to provide a direct parenchymal injury with precise control of impact diameter, velocity, depth, and dwell time ^77,78. This produces intracranial hemorrhage, cortical contusion and tissue loss, and blood-brain barrier dysfunction, which are seen in human TBI⁷². Other commonly employed open head injuries include fluid percussion injury (FPI), in which energy is transmitted to the brain via a column of fluid⁷⁹, and falling weight TBI, in which an object of a specified mass is dropped from a specified height to produce injury⁸⁰. These models have been used to study both gray and white matter pathologies following TBI, although the pathological focus is often cortex with some consideration for subcortical hippocampus⁷⁶.

In contrast to open head TBI models, “closed head” models involve the application of force directly to an intact skull, with indirect transmission of energy to the underlying brain parenchyma. Controlled cranial impact via a piston and weight drop have both been adopted for closed TBI models^73,81,82. The benefits of closed TBI models includes less anesthesia exposure due to forgoing a surgical craniotomy, as well as closer approximation to human TBI. A drawback, however, is that closed models have more heterogeneity of brain injury, also mimicking the human clinical experience. By varying the energy of the impact, this model has been used for both moderate/severe TBI, with gray and white matter injuries, and mild TBI, producing predominantly white matter injury^15,47,83.

Rotational injury models provide a distinct type of TBI that differs from the stab wound and percussive injuries. Also a closed head injury model, rotational acceleration replicates the destruction that can be seen in road accidents and sports injuries where the brain and skull move at different velocities, rupturing vasculature and causing tissue shear between brain areas of different densities⁸⁴. Prominent features of this injury model include diffuse axonal injury and sometimes subdural bleeding rather than parenchymal contusions⁸⁴. While older models require skull fixation in order to deliver a reproducible rotational force, newer methods permit free movement of the head and reproduce significant white matter injuries without gray matter involvement^85,86. As this injury favors white matter injury, cortical NG2-glia have not been the focus of much research⁸⁷.

4. THE CORTICAL NG2-GLIAL RESPONSE TO TBI

4.1. Stab Wound Injury

The largest body of work examining the cortical NG2-glia response to TBI has utilized the stab wound injury (SWI) model. While cellular proliferation after a stab wound was first described in 1970⁸⁸, the proliferative response of NG2-glia, specifically, was detailed in the 1990’s¹²: NG2 gene and protein expression as well as NG2⁺ cell density increased in areas adjacent to the injury tract, reaching a peak density at 7 days post-injury and regressing^12,89. Nevertheless, even at 30 days after injury, a small population of cells with increased NG2 decoration remained around the remnant injury site¹². Follow-up studies utilizing pulse-chase methods of following mitotic activity, found that relative to the total rate of cellular proliferation following SWI (i.e. 12- to 22-fold increase^12,13), NG2-glia comprised the largest percentage of the acute response (i.e. 1-2 days), regardless of experimental method (i.e. ~25-50% of BrdU⁺ cells^13,90–92, 82% of ³H-thymidine containing cells¹², 51% of retroviral tracer expressing cells⁷³). Following the early peak of NG2⁺ cell proliferation, other cell types (e.g. microglia, monocytes, astrocytes) predominate such that at 1-2 weeks after injury, only 20-50% of actively proliferating cells express NG2^12,13,90,91. Of note, it is currently unclear whether all cortical NG2-glia are capable of reacting to TBI equally, or if injury response is yet another property subject to NG2-glia heterogeneity. For instance, expression of G-protein coupled receptor 17 (GPR17)⁹³, which was initially thought to indicate maturation of NG2-glia from a precursor to a pre-myelinating phenotype⁹⁴, may denote a unique pool of NG2-glia with limited capability to further differentiate in the naïve animal, but maintain the capacity to expand and differentiate following a large ischemic insult⁹⁵.

On a very basic level, the proliferation of cortical NG2-glia following injury appears to play a significant role in wound healing. For instance, genetically impairing proliferation of NG2-glia leads to a larger lesion size following identical stab wounds as early as 4 days post-injury, suggesting that the response of NG2-glia is critical for acute wound closure⁵⁶. The precise mechanism by which NG2-glia affects wound healing, however, is unknown. One possibility is that production of the NG2 proteoglycan, itself, plays a significant role. The size and density of the NG2-immunoreactive “plaque” surrounding SWI appears to peak at around 7-10 days after injury, following the peak of NG2⁺ cell counts¹² and consistent with ongoing NG2 deposition at the site of the injury⁹⁶. However, several studies using various models of surgical CNS injury (most often in spinal cord) have produced disparate conclusions regarding the regenerative or inhibitory role of NG2 on the regrowth of axons^97–100, and their generalizability to cortex is not guaranteed. While no genetic NG2 knockout (KO) studies have been performed in SWI as in other TBI models (below), at the very least NG2, itself, appears to play a significant role in the polarization and migration of cortical NG2-glia towards the injury site¹⁰¹.

One potential contribution of NG2-glia to TBI recovery is to replace lost cells. The cellular fates of NG2-glia following TBI has been the subject of intense study, as few NG2-glia undergo apoptosis or necrosis following TBI^51,91,92. It is generally accepted that following TBI/SWI, NG2-glia predominantly proliferate and give rise to additional NG2-glia^51,52, with only a minority of NG2-glia expressing mature OL markers (e.g. GalC, O4), and not until weeks after injury^91,92. This is not unique to TBI, as a similar lineage restriction is seen in cortical NG2-glia after ischemic stroke as well¹⁰². While most agree that NG2-glia do not differentiate into microglia following TBI³⁹, interpretation of data supporting the potential for NG2-glia to give rise to astrocytes following injury has evolved over the years. Of note, NG2-glia-derived astrocytes are commonly referred to as Type 2 astrocytes, which differentiates them from Type 1 astrocytes derived from glial progenitors of the ventricular zone¹⁰³. The post-injury appearance of cells co-expressing NG2 and GFAP^90,92,104 or markers of immature astrocytes (e.g. nestin, vimentin)⁹⁰ initially supported an astrocytic fate for cortical NG2-glia, as can be induced in vitro⁹¹. In hindsight, these early experiments attempting to trace proliferating cells were likely complicated by not being able to account for proliferative cells lacking obvious lineage markers (e.g. an astrocyte precursor) that could contribute to the astrocyte population. Subsequent lineage mapping studies using transgenic approaches (e.g. Ng2-Cre, Olig2-Cre) were less supportive of a significant contribution of NG2-glia to post-TBI astrogliosis. Ng2-dependent lineage tracing found that the vast majority (~85%) of Ng2-derived cells remained NG2⁺ following SWI, a small population expressed mature OL markers (<4%), and a minority transiently expressed GFAP (~6%), peaking around 10 days after injury. Of note, these Ng2-derived GFAP⁺ cells comprised a small percentage of total peri-lesional reactive astrocytes (<10%), and some continued to express NG2 and were more morphologically similar to NG2-glia than astrocytes¹⁰⁵. A complementary Olig2-dependent lineage tracing strategy produced similar results, as 75-80% of cortical Olig2-derived progeny post-injury were NG2⁺, with under 5% transiently co-expressing astrocyte markers at 3 days post-injury, and the remainder expressing a mature OL marker⁵¹. When these studies were placed in context of additional fate-mapping efforts that characterized astrocyte and astrocyte precursor proliferation following injury^106,107, likely accompanied by transient expression of oligodendrocyte lineage marker Olig2¹⁰⁸, it is very unlikely that NG2-glia are a significant source of new astrocytes following SWI. Single-cell sorting has permitted a more sophisticated manner by which to characterize NG2-glia (i.e. by gene expression), including NG2-glia populations that more closely resemble oligodendrocytes and astrocytes. However, SWI does not appear to induce as strong expression of astrocyte-related markers, unlike focal ischemic injury⁴⁴. It is possible that a TBI model with a more severe injury could feature a component of focal ischemia that could induce the appearance of this population of cells.

While the lineage potential of cortical NG2-glia appears to be restricted in the normal response to SWI, several efforts have been made to uncover key molecular checkpoints that could serve as therapeutic targets to permit NG2-glia to repopulate multiple cell types lost in TBI. Olig2, a critical OL lineage transcription factor, was initially implicated as a key molecule in determining NG2-glia fate, with nuclear retention of OLIG2 in NG2-glia being associated with OL fate choice, and cytoplasmic translocation portending progression to GFAP⁺ astrocytes^91,92. However, given the astrocyte lineage mapping study above¹⁰⁵, the lack of additional studies appreciating cytoplasmic localization of OLIG2 in NG2-glia^105,108, the identification of an OLIG2-expressing population of astrocytes¹⁰⁹, and a failure to effect astrocyte proliferation following NG2-glia-specific Olig2 KO¹⁰⁵, it is less likely that NG2-glial expression of Olig2, itself, plays a determining role in shifting NG2-glia lineage potential towards astrocytes.

The bone morphogenetic protein (BMP) pathway, however, has been more consistently implicated in determining NG2-glia astrocytic lineage potential. In vitro, BMP signaling induces progression of NG2-glia to astrocytes, and BMP inhibition restricts lineage potential towards OLs^104,110. In vivo experiments may also support this hypothesis. While not a strict lineage-tracing experiment, BMP pathway dis-inhibition via anti-noggin antibodies in the setting of SWI appears to produce a more vigorous NG2-glia response and the appearance of NG2-expressing, Type 2 astrocytes¹⁰⁴. Similarly, surgical implantation of a parenchymal infusion cannula (i.e. a trauma most similar to a chronic cortical stab wound) with chronic administration of exogenous fibrinogen, thought to activate BMP signaling pathways in NG2-glia, induced cannula-adjacent NG2-derived cells to express GFAP, consistent with a shift in lineage towards astrocytes¹¹¹. This latter finding is particularly interesting as hemorrhage and extravascular deposition of fibrin are features of SWI and TBI in general. The effects of cleavage or blockade^112,113 of endogenous extravasated fibrin on in vivo NG2-glia lineage potential following TBI has not yet been explored, although at least one study supports such a link¹¹⁴.

The possibility that NG2-glia have the capacity to differentiate into neurons, both in physiologic conditions and after injury, is controversial and well-reviewed^11,115. As noted in the above lineage tracing studies, no neurons have been found as progeny of NG2-glia following in vivo SWI^51,105, casting significant doubt on the capacity for neuronal repopulation by NG2-glia in physiologic conditions. However, manipulation of certain key molecules have been reported to elicit a neuronal linage potential. Specifically, viral vector-mediated suppression of Olig2¹³ or expression of transcription factors NeuroD1¹¹⁶ or Sox2¹¹⁷ in NG2-glia were capable of inducing neuronal fates in vivo. Nevertheless, corroborative reports using complementary approaches to validate these findings in cortex are still required. As mentioned above, NG2-glia-specific Olig2 knockout failed to produce neuronal cell fates¹⁰⁵, once again introducing uncertainty in its role in altering cell fate determination. Meanwhile, targeted Sox2 overexpression did show promise in a spinal cord injury model by inducing NG2-derived neurogenesis¹¹⁸. While there is doubt that NG2-glia can repopulate lost neurons following TBI absent a molecular intervention, the potential for such a strategy should continue to motivate NG2-glia researchers to continue to identify differentiation checkpoints and further probe whether NG2-glia-derived neurons are incorporated into functional circuits following brain injury.

4.2. Open Head Injury

The second largest body of literature investigating NG2-glia following TBI is in the open head controlled cortical impact model of TBI. The fluid percussion model has also been used, but with a focus on NG2-glia in white matter injuries¹¹⁹. Similar to that seen in SWI, NG2-glia comprise the largest population of proliferative cells immediately following open TBI, with a peak at 1-7 days following injury, and a more severe injury causing a later peak^120–122. Interestingly, control mice undergoing a “sham” injury (i.e. a surgical craniotomy without impact) also demonstrate significant NG2-glia response with a proliferative cellular density similar to those having undergone impact¹²⁰, suggesting that breach of the cranial vault, despite leaving the dura intact, is traumatic enough to produce a localized proliferative response. This bears significant implications for future NG2-glia research in that study design will have to be deliberate in the selection of an appropriate “control” condition depending on the research question of interest and sensitivity of techniques to detect the cellular/molecular consequences of the “sham” injury.

As in SWI, open TBI induces significant NG2 deposition at the site of injury that persisted for up to 28 days⁴⁰. Genetic Cspg4 (i.e. Ng2) KO in open TBI models exacerbates lesion severity and functional recovery, further lending a role for NG2-glia beyond cell replacement¹²³. Of note, the Ng2 KO condition also induced a more robust astrogliosis and innate immune response, suggesting that NG2-glia, via NG2 deposition, may serve to direct the actions of other cell types neighboring the injury¹²³. Interestingly, the authors did not observe a difference in the density of PDGFRA-immunoreactive cells following injury in the Ng2 KO condition, although the presence of disparate injury severities between the two groups complicates interpretation of these data¹²³.

Open TBI models have been used to probe the engagement of specific molecular pathways by NG2-glia following TBI. One group correlated post-TBI NG2-glia proliferation with the activation of β-catenin signaling, a downstream effector of the Wnt pathway that was rarely active in the naïve state (~2% of NG2-glia)¹²⁴. Furthermore, the increased β-catenin activity may remain elevated for several days in peri-injury NG2-glia following proliferation, suggestive of prolonged activation of downstream targets¹²⁴. However, the necessity and sufficiency of β-catenin activity for precise post-injury functions, such as proliferation and/or differentiation, is unknown. Interestingly, upregulation β-catenin activity using the same reporter mice was not seen in a spinal cord injury model, suggesting a differential response between NG2-glia in cortex and versus the cord^124,125.

Another group discovered that the majority (~85%) of peri-injury NG2-glia demonstrate upregulation of nuclear factor erythroid 2-related factor (Nrf2) in a time course that mirrors the peak in NG2-glia proliferation (i.e. peak at 3 days post-injury)¹²². As Nrf2 is thought to activated by oxidative stress to initiate protective mechanisms¹²⁶, it is also elevated in several cell types following TBI, including neurons, astrocytes, and microglia¹²². While the authors briefly mentioned the existence of data in which TBI in a Nrf2 global knockout condition increased NG2-glia proliferation, it is unclear whether the result was due to a role for Nrf2 in NG2-glia, specifically.

4.3. Closed Head Injury

While closed head injury models can induce cortical NG2-glia proliferation similar to that seen in open head models, they are more commonly used to evaluate white matter NG2-glial processes post-TBI^{15,45–47,119}. Using a closed head TBI model¹²⁷, our laboratory investigated the role of the circadian clock in governing NG2-glia proliferation, both at rest and following neurotrauma¹²⁸. Interestingly, NG2-glia expressed the circadian clock gene Bmal1 both, during the time of day of peak basal proliferation as well as acutely following TBI. Additionally, utilizing a NG2-glia-specific Bmal1 conditional KO strategy, we found that Bmal1 was necessary for cortical NG2-glia proliferation, both at rest and following TBI. These data suggest that the BMAL1 transcription factor may serve as part of a common pathway by which to engage cellular proliferation, either at rest or following injury. However, while the role of Bmal1 in the mammalian circadian clock is well characterized¹²⁹, the upstream regulators and downstream effectors of Bmal1 beyond molecular clock components in NG2-glia have not yet been identified. Thus, a defined molecular pathway connecting Bmal1 to proliferation, and its interaction with the circadian clock and TBI is unknown.

5. DISCUSSION AND FUTURE DIRECTIONS

There have been many descriptions of NG2-glia proliferation and differentiation following TBI, however, researchers have only begun to identify the molecular mechanisms regulating these processes, with most focusing on individual molecular candidates (Figure 2). Given the acceleration in cell-type-specific next generation molecular profiling, we anticipate that future studies will be better suited to elucidate entire molecular pathways responsible for the NG2-glial post-TBI response, which could provide potential targets for therapeutic manipulations. We also expect that these studies will divine which pathways bear some specificity for NG2-glia, as several of the molecular candidates discussed above are fairly ubiquitous (Wnt/ β-catenin, Nrf2, Bmal1). Likewise, as we continue to take advantage of studies with single-cell resolution, we expect to completely redefine the NG2-glia population beyond the use of markers such as NG2 and PDGFRA, much like has been recently done for microglia¹³⁰, as there is increasing evidence that even NG2-glia within a particular neuroanatomic location may be functionally distinct^44,95.

Figure 2. Molecular pathways controlling cortical NG2-glia proliferation and differentiation.

Cortical NG2-glia respond to neurotrauma with robust proliferation and differentiation. (A) Reactive NG2-glia (green) polarize and migrate towards the site of TBI, while NG2-glia remote from the injury (magenta) lack any apparent morphologic changes. (B) In the acute response to TBI, reactive cortical NG2-glia (green) most commonly proliferate, leading to additional NG2-glia (large black arrow). A smaller population of NG2-glia undergo differentiation to mature oligodendrocytes (small black arrow). While data from stab wound injury models suggest that only few NG2-glia may be able to differentiate into Type 2 astrocytes or “astrocyte-like NG2-glia” (gray arrow), it is possible that larger lesions with ischemia or more significant blood brain barrier compromise may promote an astrocytic lineage potential through activation of BMP signaling (e.g. fibrinogen). Although NG2-glia have not been observed to differentiate into neurons under physiologic conditions in vivo (yellow arrow), virus-mediated expression of specific transcription factors (e.g. Sox2, NeuroD1) may be able to induce such a fate following TBI.

Created with BioRender.com

Another knowledge gap to be addressed is the change in functions of NG2-glia over time following injury. While the majority of studies appear to be focused on the immediate response of NG2-glia within the first few weeks after injury (Supplemental Table 1), there is evidence of morphological changes approaching a month after neurotrauma^12,14. Furthermore, literature from static white matter injury (e.g. HIE, TBI) suggest that there is a prolonged NG2-glial response in the chronic phase following a CNS injury^24,131; the functions of cortical NG2-glia, and how they may contribute to brain healing, in the following months after TBI are entirely unknown.

Beyond the scientific advances that will improve resolution of NG2-glia molecular pathways, much attention moving forward should also be focused on the choice of TBI model. As shown here, investigations into the functions of NG2-glia following TBI are most complete in SWI models. This model, while offering the best experimental reproducibility, is limited in generalizability due to its inherent differences compared to the human condition and relatively mild degree of injury. Clinical TBI features cellular destruction, immune interaction, and blood vessel compromise, all of which contribute to focal ischemia, hemorrhage, and blood brain barrier dysfunction that may be more severe than in a focal stab wound. Even the use of a larger “stab” wound versus a smaller “punctate” wound appears to produce a more robust NG2-glial polarization and migration relative to the injury site⁵⁶. Thus, there is increasing support for TBI researchers aiming to bridge the gap between bench and bedside to use multiple standardized experimental TBI models (including injuries in gyrencephalic species), with therapeutic “rescue” being defined as positive changes in several peripheral and central biomarkers and clinically relevant behavioral testing^132,133. While the translation of molecular mechanisms to clinical therapies has been slow, if not impossible¹³⁴, we are hopeful that NG2-glia present a unique cell type with the potential to enhance cellular recovery following TBI.

9. Data Availability:

No new data are included in this review article.