Damage Mechanisms to Oligodendrocytes and White Matter in Central Nervous System Injury: The Australian Context

Oligodendroglia diversity

OPCs derived from different embroyonic regions produce oligodendrocytes that preferentially populate different regions across the adult CNS. ⁵⁴ Indeed, oligodendrocytes originating from the ventral region of a developing mouse brain mainly populate the anterior commissure, pre-optic tract, and the lateral olfactory tract, whereas their dorsally derived counterparts predominatly localize to the cortex and corpus callosum. ^48,54 Further, dorsally and ventrally derived oligodendrocytes preferentially myelinate dorsal and ventral regions of the adult CNS, respectively, ⁴⁸ despite having a similar ability to generate mature oligodendrocytes. ⁵⁵ Characteristic differences also exist between OPCs in white and gray matter, with the former having a fourfold faster rate of proliferation and a higher proclivity for maturing into myelinating oligodendrocytes. ^44,52,56

With respect to transcriptional heterogeneity, single-cell RNA profiling of oligodendroglia in different regions of the mouse adult CNS identifed 12 transcriptionally distinct subpopulations, comprising OPCs, differentiation-committed OPCs, newly formed oligodendrocytes (NFOL), myelin-forming oligodendrocytes (MFOL), and mature oligodendrocytes (MOL). ⁵⁷ In contrast to OPCs, differentiation-committed OPCs lack PDGFRα and NG2 but express genes such as sex-determining region y-box (SOX) 6, Neu4 and Bmp4 that keep oligodendrocytes undifferentiated. NFOL are distiguishable from MFOL by their expression of genes associated with early stages of differentiation (such as Tcf7l2 and contactin-associated-protein [Caspr]) and downregulation of genes involved in myelin formation, including myelin oligodendrocyte protein (MOG), proteolipid protein 1 (PLP1), and oplalin. MOL express genes associated with late oligodendrocyte differentiation and myelin formation simultaneously. Whereas OPCs and differentiation-committed OPCs were found to be transcriptionally homogenous, two populations of NFOL (NFOL1 and NFOL2) and MFOL (MFOL1 and MFOL2) and six populations of MOL (MOL1 to MOL6) with distinct gene expression profiles have been identified. ⁵⁵

These subpopulations differed in spatial distribution within the regions of the CNS analyzed (i.e., amygdala, corpus callosum, cortex, zona incerta, dentate gyrus, striatum, dorsal horn, hippocampus, hypothalamus, and substantia nigra/ventral tegmental area; SN/VTA). NFOL1 and NFOL2 were found in every region investigated except the corpus callosum, where NFOL2 was absent. MOL subpopulations varied in region specificity, with some populations, such as MOL5, being present throughout the regions, whereas others were enriched in particular regions; for example, MOL2 were abundant in the spinal cord but almost absent in the brain. ⁵⁵

Although the functional implications of the transcriptional heterogeneity and spatial preferences of oligodendroglia are still unclear, a recent study suggested that MOL subpopulations may fulfil distinct regenerative functions, as MOL2 were depleted at the site of SCI and, in contrast, MOL5/6 increased their contribution to lineage repopulation and thus remyelination. ⁵⁵ The development of different oligodendroglia subpopulations and their progression from OPCs to mature myelinating oligodendrocytes is tightly regulated by complex interactions between several transcription, post-transcription, translation, and post-translation factors.

Molecular control of oligodendroglia development

During embryonic development of the CNS, distinct molecular mechanisms control the production of dorsally and ventrally derived OPCs. Sonic hedgehog (SHH) signaling induces ventral OPC specification by driving NKX6.1, NKX6.2, NKX2.2, and Olig2 transcription in neural stem cells. ⁵⁸ By contrast, dorsal OPC generation is regulated in a SHH-independent manner by Dbx1 and Ascl1 transcription factors and fibroblast growth factor and bone morphogenetic protein (BMP) signaling. ⁴⁸ As OPCs migrate from their birthplace to populate the CNS, a complex network of transcriptional and translational changes coordinate to mediate OPC differentiation. Calcineurin, a calcuim-dependent phosphatase, was recently identifed as a key initiator of OPC differentiation. ⁵⁹ Triggered by an increase in intracellular Ca²⁺ levels, calcineurin dephosphorylates the nuclear factor of activated T cells, cytoplasmic 2 (NFATC2) protein, resulting in its activation. In turn, activated NFATC2 interacts with SOX 10 transcription factor to allow Olig2 and NKX2.2 co-expression, which is a prerequisite for OPC differentiation. ^60,61 Numerous transcription factors have been reported to play a role in oligodendrocyte differentiation over the years, including nuclear receptors such as retinoid X receptors, vitamin D receptors, and peroxisome proliferator-activated receptors that contribute to oligodendrocyte formation and myelination. ⁶² However, Olig2, SOX10, NKX2.2, zincfinger protein 24 (ZFP24), and MyRF are widely recognized as the main regulators of full oligodendrocyte differentiation and myelination. ⁶² SOX10 plays a central role in this process, binding and directly activating several genes involved in oligodendrocyte maturation, including MyRF, which mediates terminal differentiation of immature premyelinating oligodendrocytes. ^60,61 In addition, SOX10 induces expression of NKX2.2 and Olig2, the latter of which activates ZFP24 to generate myelin transcripts. ⁶⁰

Genes involved in lipid and cholesterol synthesis are also associated with myelination of mature oligodendrocytes. ⁶³ The transcription factors sterol regulatory element binding proteins (SREBPs) have been reported to control the supply of fatty acids for myelin formation through the mammalian target of rapamycin (mTOR) signaling pathway. ^63,64

Oligodendrocyte development can also be regulated epigenetically. At the OPC stage of lineage progression, the transcription factors and histone-modifying enzymes that respond to extracellular stimuli are mainly inhibitory, preventing differentiation. ⁶⁵ Long noncoding RNAs (lncRNAs) are approximately 200 nucleotides long, and regulate gene expression but lack protein-coding potential. ⁶⁶ Oligodendrocyte-specific lncRNAs such as lncOL1 and lnc158 can downregulate inhibitors of histone deacetylase activity and heterochromatin formation to promote expression of oligodendrocyte differentiation genes. ^67,68 MicroRNAs (miRNAs) are another subgroup of ncRNA that have been shown to regulate myelination in oligodendrocytes. They are short nucleotides (20–25 nucleotides long) generated by enzymatic cleavage of long transcripts. ⁶⁹ Numerous miRNAs with a role in oligodendrocyte development have been identified and reviewed in the literature. ⁶² Of note, a recent study showed that oligodendrocyte-specific miRNA-219 and miRNA-338 collaborate to downregulate negative regulators of oligodendrocyte myelination including the transcription factors SOX6 and ETS variant 5. ⁷⁰ In addition, miRNA-219 represses expression of proteins such as PDGFRα that maintain OPCs in their progenitor state, thereby promoting oligodendrocyte differentiation. ⁷⁰

After oligodendrocytes are generated, myelination is triggered by the downregulation of precursor proteins such as PDGFRα and transcription of MyRF and its target proteins, which are essential for myelin formation, including MBP, PLP1, myelin-associated glycoprotein (MAG), cyclic nucleotide phosphodiesterase (CNP), and MOG. ⁷¹ Myelinating oligodendrocytes use their expanded network of processes to make stable contact with neighboring axons, enwrapping segments of the neuron body with myelin sheaths. ^72,73 The structure and function of myelin is central to white matter composition and physiology as white matter is mainly made up of myelinated axons.

Myelin structure and function

Myelin has a very high lipid content, comprising phospholipids, galactolipids, and plasmalogens ⁷⁴ that have an affinity for, and bind, myelin-specific proteins. ⁷² The myelin sheath consists of compacted double bilayers of myelin, with myelin proteins, particularly MBP, acting as an adhesive between the bilayer membranes. ⁷⁵ Once matured, myelin can consist of up to 160 layers of membranous, compacted lamellae that can be visualized as interperiodic lines using electron microscopy. This myelin sheath allows for saltatory conduction of action potentials and refined neural signaling, ⁷⁶ with myelinated axons conducting signals at a speed 20 to 100 times that of unmyelinated axons of similar diameter. It has long been known that the thickness of the myelin sheath is proportional to the diameter of the axon ⁷⁷ ; however, it was recently discovered that both axonal diameter and the corresponding myelin thickness can vary along the length of a single axon. ⁷⁸

Myelin has a highly segmented structure to facilitate neuronal signaling. ⁷⁹ The myelin sheath produced by mature oligodendrocytes along an axon comprises multiple myelin segments, called internodes, separated by unmyelinated regions called nodes. ⁸⁰ Oligodendrocytes myelinate axons of all diameters above ∼0.4 μm, ⁷² and a single oligodendrocyte can provide up to 50 internodes of myelin on different axons. ⁸¹ The length of internodes and processes of an oligodendrocyte limits its ability to myelinate the same axon twice. ⁵¹

The spaces between internodes are referred to as nodes of Ranvier. ⁸² Localized at the node of Ranvier are specialized sodium channels that facilitate and propagate action potential signaling. ⁸³ Thermosensitive and mechanosensitive K2P potassium channels (TREK-1 and TRAAK) are clustered at the nodes of Ranvier and are the principal drivers of rapid action potential repolarization. ⁸⁴ Directly adjacent to the node is the paranode, which consists of loops of myelin lamellae that are anchored to axons via the paranodal junction, which comprises contactin, neurofascin 155, and Caspr. ⁸⁵ The paranode has multi-faceted roles in maintaining saltatory conduction; including (1) segregating nodal sodium and juxtaparanodal potassium channels; (2) stabilizing and attaching the myelin sheath to the axon to prevent myelin damage from stretch stressors; and (3) allowing only specific nutrients to diffuse into and out of the internodal peri-axonal space. ⁸⁶ The juxtaparanode, which lies between the paranode and the internode, contains a high concentration of voltage-gated potassium channels, ⁸³ although the function of these specific to their juxtaparanode location is currently unclear. ⁸²

Neuron-oligodendrocyte interaction is integral for axonal integrity. In addition to their well-established function of ensuring fast action potential propagation and axonal insulation, oligodendrocytes supply neurons with nutrients. ⁸⁷ Oligodendrocytes transfer energy metabolites including pyruvate and lactate to neurons, through cytoplasmic channels and monocarboxylate transporters, allowing fast adenosine tri-phospate (ATP) synthesis. Further, oligodendrocytes contribute to the maintenance of axon integrity and survival of neurons by producing neuroprotective and trophic factors such as ciliary neurotrophic factor (CNTF), insulin-like growth factor-1 (IGF-1), and glial cell line-derived neurotrophic factor. ⁸⁸ Indeed, partial ablation of oligodendrocytes causes axonal degeneration, even in the absence of widespread demyelination. ⁸⁹ Loss of oligodendrocyte-specific proteins such as PLP and CNPase results in axonal swelling and degeneration, even if myelin structure remains normal. ^90,91 These seminal studies confirm that oligodendrocytes play multiple roles in neuronal maintenance. A further means of interaction is the release of trophic factor exosomes from oligodendrocytes, which are taken up by neurons and improve their metabolic activity. ⁹² In addition, oligodendrocyte-neuron interactions are adaptive and plastic, with neural signaling promoting oligodendrogenesis and increased myelin thickness. ^{93 –96}

Other than acting as a pool of progenitors for oligodendrocytes, OPCs play multi-faceted roles in the CNS. Ablation of OPCs results in excessive inflammation, suggesting OPCs may suppress neuroinflammation in a healthy system. ⁹⁷ It has recently been discovered that peri-vascular OPCs wrap their processes around blood vessels in the brain and may be partly responsible for regulating and modulating the neurovascular unit. ⁹⁸

Specifically, OPCs play a role in maintaining blood–brain barrier (BBB) integrity via communication with cerebral endothelial cells, ⁹⁹ including via transforming growth factor (TGF)-β signaling. OPCs located around blood vessels secrete TGF-β1, which activates the mitogen-activated protein kinase (MEK)/extracellular signal-regulated kinases (ERK) pathway in cerebral endothelial cells, increasing expression of tight-junction proteins and protecting BBB integrity. ¹⁰⁰ In turn, cerebral endothelial cells have been found to release growth factors such as brain-derived neurotrophic factor and basic fibroblast growth factor, which promote the survival and proliferation of OPCs, both of which are significantly decreased following stroke. ^101,102 This trophic coupling between the cerebral endothelium and neighboring OPCs has been dubbed the oligovascular niche. ¹⁰² In vitro studies have also shown that astrocytes can support OPCs, via MEK/ERK and PI3K/Akt signaling, ¹⁰³ and OPCs and peri-cytes could potentially secrete factors to support each other. ¹⁰⁴ Further, within the developing brain, OPCs secrete hypoxia-induced factor to promote angiogenesis of the CNS vasculature. ¹⁰⁵

Models of CNS injury

For the mechanisms of white matter degeneration following CNS injury to be studied, it is important to utilize reliable animal models of injury. Animal models can be employed to replicate human TBI and SCI, and experimental models can be amended to study specific facets of heterogeneous pathologies, allowing mechanisms to be studied in a rigorous and controlled manner. ^115,116 Rodent models of injury are typically employed by researchers as they are cost and time effective; however, many translational challenges exist due to inter-species differences in neurometabolism, lifespan, and anatomy. ¹¹⁶ Animal models of SCI include spinal contusion and compression, which best simulate the biomechanics and pathology of human injury. ¹¹⁵ Most SCI models use a thoracic injury site as opposed to cervical, as the thoracic SCI model tends to be more reproducible, has a lower risk of mortality, and permits the study of secondary white matter deficits. ¹¹⁵ Prominent models of TBI include: blast injury models, fluid-percussion injury (FPI), controlled cortical impact (CCI) models, and closed-head weight-drop models. ^{117 –119} Additionally, TAI can be induced in mice using closed-skull stereotaxic impactors delivering single or repeated mTBI to study microstructural changes with electron microscopy and observe long-term behavioral changes. ^{120 –122}

Although all of the aforementioned models robustly elicit an inflammatory response and white matter injury, FPI and CCI models (which require the animal to be stationary) fail to mimic the human head kinematics following TBI. ^15,116 The closed-head impact model of engineered rotational acceleration (CHIMERA) is a recent innovation that allows highly reproducible and accurate mTBI delivery. ¹²³ The CHIMERA model induces DAI, behavioral changes, and energy dose-dependent increases in axonal damage. ¹²⁴ Closed-head weight-drop models have also been employed in recent years, as they approximate the angular and rotational forces of human TBI and allow study of repeated TBI. ^15,116,125 White matter reductions and demyelination are observed in the corpus callosum in the closed-head weight-drop model; therefore it remains an appropriate method to replicate human pathology. ¹⁵ A caveat of the rodent model is that there is proportionally less cerebrospinal fluid (CSF) as compared with humans; therefore rotational effects are not necessarily comparable. ^126,127 Angular acceleration forces are reduced in rodent brains due to their small physical size; therefore white matter injury may be reduced and less clinically relevant. Large animals have a higher white-to-gray matter ratio due to their gyrencephalic structure, and therefore are better suited for translational validation of novel discoveries in rodent models. ¹²⁷

In many models of neurotrauma, it is challenging to distinguish between primary and secondary damage, as there is often no clearly defined margin between tissue damaged by the initial insult and spared tissue. ¹²⁸ Partial optic nerve transection is an established and useful model for investigating secondary degeneration of a white matter tract that features a primary injury of a penetrating nature that does not mechanically damage adjacent tissue. ¹²⁹ In the model, the right dorsal optic nerve is partially transected, which allows for spatial separation between primary and secondary injuries. ¹²⁹ The partial optic nerve transection model results in transient disruption of the BBB, which can be a useful feature for proof of principle investigations into secondary degeneration of white matter with a transiently open BBB. ¹³⁰ The optic nerve stretch model allows for the study of wallerian degeneration often seen in CTE. ¹³¹ The stretch model and the transection model differ as the stretch model analyzes distal injured axons, whereas in the partial transection model uninjured axons ventral to the injury site are analyzed for secondary degeneration.

Detecting lesions and volume changes

MRI is a reliable tool that can provide high-resolution images allowing for the detection of white matter lesions, detected as changes in iron and fluids due to hemorrhages and swelling in the brain. In a blinded study in the acute phase of TBI or minor stroke (<48 h post-injury), clinicians were able to accurately discriminate TBI from acute minor stroke: MRI-facilitated detection of vascular abnormalities in 54% of patients with TBI, whereas only 14% of patients had positive vascular findings using CT. ¹⁴² Hyperintense white matter lesions can be detected following TBI using T2-weighted fluid-attenuated inversion recovery (FLAIR) sequences. FLAIR nullifies body fluids during acquisition, therefore suppressing CSF effects on image contrast and allowing peri-ventricular hyperintensities and meningeal focal enhancements common in TBI to be detected with greater accuracy. ^142,143 Presentations of lesions following TBI are heterogeneous. Although the number and volume of lesions correlates with adverse outcomes, ¹⁴⁴ the presence of white matter hyperintensities and microhemorrhagic lesions in mTBI are uncommon and do not necessarily predict persisting symptoms. ^145,146

However, in cases of moderate and severe TBI, microhemorrhages are an independent predictor of disability (GCS-extended score ≤6; odds ratio = 2.5) in the absence of axonal injury. Therefore, caution should be taken when interpreting lesion data, as microhemorrhages have a significant role in TBI. ¹⁴⁷ Additionally, T2-weighted MRI may be limited in its utility as lesions do not often appear in the acute phase following mTBI, but are more often detected at 6 month follow-up in cases of persisting symptoms. ¹⁴⁸ However, in cases of severe TBI, lesions referred to by the authors as TAI in FLAIR sequences in the corpus callosum and the brainstem are predictors of reduced functional outcomes at 12 month follow-up. ¹⁴⁹ Nevertheless, lesion presence alone does not provide sufficient prognostic clinical value.

Brain volume changes provide insight into gross structural changes to white matter in the brain. Reductions in cortical thickness provide an indication of widespread atrophy in patients with severe TBI and these changes are associated with deficits in declarative memory and auditory verbal learning. ¹⁵⁰ Cortical thinning is exacerbated in patients with repeated mTBI and veterans exposed to blast injuries following mTBI. ^151,152 Despite cortical thickness reductions often taking place in tandem with reductions in white matter integrity, ¹⁵⁰ changes to white matter integrity can take place independent of cortical thinning. ¹⁵³ Widespread atrophy is also inferred by total brain volume reductions, with patients with severe TBI having on average a 4% reduction in brain volume at 12 months post-injury, indicative of long-term TAI. ¹⁵⁴ Additionally, atrophy is greatest in white matter compared with gray matter. ^155,156 Regional brain volume reductions have been observed in the white matter of the anterior cingulate, ¹⁵⁶ brainstem, thalamus, and corpus callosum. ¹⁵⁴ In SCI, white matter volume loss is prevalent in the corticospinal tract, although gray matter atrophy appears to be more robustly evident compared with heathy controls. ¹⁵⁷ Additionally, ventricular enlargement and CSF volume increases may serve as a surrogate marker of ascending neurodegeneration following SCI. ¹⁵⁸ Taken together, it is clear that brain volume changes provide clinically relevant insight into longitudinal atrophy of both white and gray matter.

DTI connectivity changes

White matter fasciculi can be accurately mapped and characterized using diffusion tensor imaging (DTI) techniques. DTI utilizes the anisotropic nature of water displacement in white matter to derive values such as fractional anisotropy (FA), which characterizes diffusion anisotropy differences between intensity limits of zero and one. In highly organized white matter tracts, FA values are high (∼0.8) due to diffusion approaching unity along white matter tracts and diffusion in all directions perpendicular to tracts. In gray matter, FA values generally vary (∼0.10–0.25) depending on the structure of the region concerned. ¹⁵⁹ In CSF, FA values approach zero as water diffusivity is equal in all directions. ¹⁶⁰ Mean diffusivity (MD) is another common measurement and quantifies mean diffusion in each direction as opposed to FA, which quantifies the total magnitude of diffusion along axonal fibers. FA is calculated as a ratio of axial diffusivity (AD) and radial diffusivity (RD), which describe the magnitude of water diffusion parallel and perpendicular to the tract, respectively. ¹⁶¹ In cervical SCI, mean FA values are reduced at the site of injury and mean MD values are increased, compared with healthy controls. The decrease in FA value may reflect a decreased degree of myelination in the tracts or increased cell density, whereas higher MD may be indicative of white matter tract disorganization or fluid edema. ¹⁶²

Diffusion basis spectrum imaging (DBSI) is a novel technique that shows promise in elucidating underlying pathology behind diffusion metric changes. ¹⁶³ DBSI allows quantification of white matter tract density, as well as residual axon and myelin integrity and axonal volume while accounting for vasogenic edema. The DBSI technique has been validated in a traumatic SCI study in mice, wherein co-existing pathologies were identified non-invasively and then correlated with post-mortem histology markers. ¹⁶³ Clinically, a decrease in FA values is also seen at the thoracic level below the injury site, ¹⁶⁴ and persists in ascending tracts into cerebral white matter tracts, with reduced FA and increased MD in the corpus callosum, corona radiata, cingulate gyrus, and the thalamic radiation years post-injury. ¹⁶⁵ White matter disruption is worse in patients with cervical compared with thoracic injury, and time since injury is negatively correlated with FA values in some brain regions. ¹⁶⁵ These DTI changes are consistent with wallerian degeneration, with discrete white matter injury in the spinal cord causing demyelination and structural disorganization distal to the injury site. ^106,164,165

Unlike SCI, DTI findings from TBI studies lack clinical consensus and are far more heterogenous. ¹⁶¹ An extensive review by Asken and colleagues ¹⁶⁶ postulates that variability in the changes in FA and MD values after injury are attributable to different characteristics of the study populations and comparative control groups. ^166,167 In the acute phase of mTBI, FA values increase and MD values decrease in the days following injury. Specifically, FA values have been seen to increase in the corpus callosum, ¹⁶⁸ fimbria, and the internal capsule. ¹⁶⁹ These FA and MD changes are believed to be due to axonal cytotoxic edema in the initial phases of white matter degeneration following injury. ^{168 –170}

Axonal swelling can occur due to interruption of axonal transport and subsequent accumulation of transported materials, or it can manifest in axonal bulbs following complete axonal disconnection. ¹⁷¹ In a rodent CCI study, acutely increased FA in the genu corpus callosum was accompanied by decreased RD and MD, with no changes in AD observed. The changes in FA observed using DTI were unable to be detected by anatomical MRI or hematoxylin and eosin staining. ¹⁶⁸ FA decreases in the contralateral cortex, thalamus, and hippocampus at 7 days have also been observed in a recent Australian study following CCI in rodents, validated by cross-sectional histopathology. ¹⁷² Accompanying acute injury in rodents, increased FA values have been associated with increased expression of tumor necrosis factor α (TNF-α) and a reduction in MBP staining. This supports the hypothesis that increased FA values immediately following injury are a result of cytotoxic processes affecting myelin integrity and therefore diffusivity along white matter tracts. ¹⁶⁹ Wright and associates recently observed similar changes in DTI measures associated with changes to telomere length. Repeated mTBI resulted in reduced telomere lengths and these changes were associated with reductions in AD and RD. ¹⁷³

An inverse pattern is observed in mTBI as time passes, wherein FA values decrease and MD values increase. ^174,175 Multiple injuries in animal models exacerbate changes in diffusivity metrics and are also associated with hypoconnectivity in the hippocampus, brainstem, and cerebellum. Interestingly, rats exposed to a single mTBI display hyperconnectivity between brain regions, suggesting an adaptive response to injury. ¹⁷⁵ A similar finding was observed following single CCI, wherein the brain was connectively promiscuous post-injury in the thalamus and cortex, perhaps reflecting circuit reorganization. ¹⁷⁶ Following repeated mTBI in mice, no changes in FA are observed acutely, but after 6 weeks FA is significantly reduced in the corpus callosum compared with sham animals. This corpus callosum FA reduction occurred in tandem with microglial activation detected by post-mortem immunohistochemistry, with activation peaking at 7 days post-injury. ¹⁷⁴

Whereas DTI can be used to detect chronic white matter aberrations, the many changes in the tissue and cells within voxels may contribute to changes in DTI measures in ways not yet fully understood. The temporal inversion of FA values is reflected in the clinical findings of DTI. A longitudinal study of DTI metrics found FA values decreased from the acute phase following mTBI (<7 day) to the subacute phase (8 days to rehabilitation discharge). The variability of FA values in the acute phase means that DTI is more likely to be useful as a prognostic tool subacutely. ¹⁷⁷ MD values are higher in athletes who have received multiple concussions compared with those who have had single or no concussions, in the absence of any major neuroanatomical differences. ¹⁷⁸

In severe chronic TBI patients, increased FA values in the splenium of the corpus callosum have been associated with better performance in social inference tests. Despite a number of gray matter abnormalities, the changes in the corpus callosum were those most implicated in functional deficits in these TBI patients. ¹⁷⁹ In combat veterans who suffered TBI while deployed, those with higher FA values and lower MD values have a greater likelihood of returning to work post-deployment. ¹⁸⁰ In terms of long-term prognosis, higher FA values in the corpus callosum in the acute phase following injury are associated with higher disability rating scores. Whereas in the subacute period, higher FA values in the splenium correlate with lower disability rating scores. ¹⁷⁷

Some brain regions, such as the internal capsule and the fronto-occipital fasciculus, show recovery from a transient reduction in FA values. Increased FA values at 1 month follow-up are positively associated with better performance on cognitive information processing speed at initial assessment. However, continued loss of integrity is observed in the corpus callosum, forceps major, and anterior corona radiata at 3 months, and these regions may serve as an objective biomarker of pathology that can predict persisting post-concussion symptoms (PPCS). ¹⁸¹ Hirad and co-workers ¹⁸² used accelerometers and DTI imaging on college football players and tracked head impacts across a competitive season in conjunction with an mTBI cohort of patients. FA in the midbrain was reduced post-season compared with pre-season, as was the case for the right midbrain in the mTBI patients. Reductions in midbrain white matter integrity in the football players were related to the amount of rotational acceleration sustained as opposed to linear acceleration, ¹⁸² and this is consistent with previous biomechanical models of TBI. ^109,110 Serum-based tau, a marker of axonal injury and BBB integrity, was inversely correlated with midbrain FA values in the mTBI patients. ¹⁸²

FA and MD metrics can also be used to add structural insight to neuropsychological outcomes in TBI patients. In a recent Australian study, ¹⁸³ patients with severe TBI had widespread white matter pathology in the corpus callosum, longitudinal fasiculi, corticospinal, brainstem, and cerebellar tracts. Diffuse white matter integrity disruption in these patients, particularly changes in the thalamus and fornix, was associated with social information processing deficits. ¹⁸³ Diffusion tensor techniques can also be applied to derive individual patient structural connectomes, in which cognitive associations can be explored. Mishra and associates ¹⁸⁴ created a structural connectome using average FA values and brain atlas-defined nodes to assess differences between cognitively impaired and non-impaired professional fighters. White matter reorganization in the hippocampus, precuneus, and insula was detected due to repeated head injury in fighters exhibiting cognitive decline. ¹⁸⁴ Recent work ¹⁸⁵ in white matter plasticity used DTI techniques as well as myelin water fraction and longitudinal relaxation rate R₁ to assess subtle changes in myelination and connectivity in healthy subjects following working memory training. The study highlighted the sensitivity of the R₁ measure in detecting myelin remodeling following working memory training, which occurred in conjunction with changes in FA and RD parameters. ¹⁸⁵ Given R₁ values may be more specific in detecting changes in oligodendrocytes. ¹⁸⁶ Future studies could employ this measure to assess white matter regeneration following cognitive rehabilitation in TBI patients.

Novel methods for increased sensitivity

The diffusion tensor model faces a number of limitations that restrict its sensitivity in detecting microstructural changes in white matter pathology. One of these limitations lies in the fact that DTI assumes a single fiber orientation and is confounded at regions of “crossing” fibers. This can be overcome by analyzing apparent fiber density (AFD), utilizing a DWI post-processing technique known as constrained spherical deconvolution. In a recent rodent FPI study conducted by Wright and colleagues, DTI and AFD measures were both able to detect white matter changes in the corpus callosum: a uniformly orientated tract. However, DTI was unable to detect any changes in the corticospinal tract, whereas there was an observed reduction in AFD in the same animals. ¹⁶⁷ Tract weighted imaging (TWI) is a novel technique that utilizes constrained spherical deconvolution to better characterize the fiber orientation distribution in regions of interest, allowing characterization of tract density, curvature, and length as additional measures that may be more sensitive to white matter pathology. ¹⁸⁷ Using TWI in an awake closed-head model of injury, Australian researchers ¹⁸⁸ found that tract density is significantly increased in the fimbria and contralateral external capsule following injury. Additionally, TWI detected changes between sham and TBI group at day 1 post-injury, whereas DTI metrics only detected changes at 7 days. ¹⁸⁸

Another inherent limitation of DTI is that it applies a Gaussian distribution model to tissue using a normative distribution. In reality, the complexity of brain tissue means that water diffusion deviates significantly from this pattern. ¹⁸⁹ As an example, when oligodendrocyte apoptosis occurs in combination with vasogenic edema, the result is enhanced diffusivity across axons. This pathological state results in a discrepancy between RD and histological outcomes. ¹⁹⁰

Diffusion kurtosis imaging (DKI) is a non-Gaussian distribution measure that accounts for tissue heterogeneity and provides improved characterization of white matter integrity. Common DKI parameters include: mean kurtosis (MK) and mean kurtosis tensor (MKT), which reflect the average diffusion in all directions; and axial kurtosis (AK) and radial kurtosis (RK), which quantify diffusion along the axial or radial direction of the diffusion ellipsoid, respectively. ¹⁸⁹ In a rat CCI study utilizing both DTI and DKI techniques in vivo, animals were imaged acutely (>2 h) and subacutely (7 days) post-injury. ¹⁹¹ In the acute phase, both DTI and DKI parameters detected differences in the injury versus sham animals. FA increased in the hippocampus and the ipsilateral cortex and MD decreased across all regions. However, at subacute testing most DTI changes had returned to baseline with the exception being a reduction in FA in the ipsilateral cortex. DKI measures were able to distinguish changes in both white and gray matter microstructure at the subacute period, with an increase in MK observed. ¹⁹¹ In addition to being more sensitive than DTI metrics in this study, kurtosis abnormalities scaled inversely with distance from impact site, therefore providing some discriminatory power to injury severity. ¹⁹¹ However, at longer time-points post-injury (42 days), it was found that DKI metrics did not provide additional sensitivity for detection of abnormalities in the corpus callosum compared with DTI metrics. ¹⁷⁴

Similar to DTI metrics, a time-dependent biphasic response has been observed in MK values following injury, whereby MK values initially decrease following injury and later reverse 7 days following injury. This biphasic change in MK values is believed to represent a shift from cytotoxic to vasogenic edema. ¹⁵⁹ Another confounding factor pertains to astrocyte activity following injury, as increases in FA can be falsely interpreted as remyelinating processes when in reality they reflect glial scar formation. ¹⁹² Increases in MK have been associated with astrogliosis, ¹⁹¹ and occur in tandem with higher density of microglia and astrocytes. ¹⁹³ A recent Australian study ¹⁹⁴ used a longitudinal rodent weight-drop design and assessed both DTI and DKI metrics. The study commended the sensitivity of both diffusion metrics; however, it highlighted that kurtosis measures were highly sensitive to changes in astrocyte markers at 3 months. Additionally, axial measures were not changed to the same degree as RK and RD metrics in the injury group. Therefore it is postulated that axonal integrity is maintained and the radial diffusivity changes may reflect cellular aggregations from glia, although these changes may also be due to demyelination. ¹⁹⁴ However, a recent clinical mTBI study found that AK values were positively correlated with working memory deficits, indicating axonal perturbations. ¹⁹⁵ Previous rodent work has also found correlations between DKI parameters and neuron loss, microglia activation, and myelin disruption. In addition, changes in MK values in the ipsilateral cortex and hippocampus are associated with deficits in cognitive function. ¹⁹⁶

Advances in diffusion imaging techniques allow neuronal projections, known as neurites, to be imaged in vivo. The model used is known as neurite orientation dispersion and density imaging (NODDI), and estimates the microstructural complexity including dendrites and axons. ¹⁹⁷ When NODDI is employed in conjunction with traditional DTI metrics, greater insight into the pathophysiological underpinnings of injury can be achieved. Repeated mTBI in martial arts athletes results in decreased FA, an increase in orientation dispersion index (ODI), and increased intracellular volume fraction (V_IC), therefore inferring a potential role of edema in traumatic injury. ¹⁹⁸ In concussed athletes, increased FA and decreased MD was detected alongside increasedV_IC and decreased ODI. The increase in neurite density and coherence of neurites in the white matter of sports-concussed athletes likely reflect an adaptive reparative response to injury, especially given the lack of functional impairment present. ¹⁹⁹ In contrast, in a study conducted in athletes who suffered mTBI, reduced FA occurred concurrently with a reduction in V_IC, in the absence of any significant changes in ODI at 7 days post-injury. These changes may be associated with glial-mediated edema and the initiation of a neuroinflammatory response to injury. In athletes scanned after being given medical clearance to return to play, ODI was increased in individuals with elevated symptoms at initial consultation. Therefore, the change in neurite geometry may reflect a longitudinal response to injury, and may be a sensitive measure of clinically asymptomatic neuroplastic changes in the brain. ²⁰⁰ A study comparing patients with mTBI with orthopedic trauma controls observed decreased FA and increased MD in anterior tracts at 2 weeks post-injury, with a concomitant elevated free water fraction (FISO) in white matter regions indicative of vasogenic edema. However, at 6-month follow-up, no significant differences were detected in DTI metrics, despite a longitudinal decrease in white matter neurite density on NODDI. Additionally, associations were found between ODI and FISO and rate and magnitude of functional improvement within the mTBI group. ²⁰¹ In summary, NODDI augments the sensitivity of existing DTI metrics and further studies are warranted to explore its prognostic and research value.

Resting state and connectivity changes

Patients suffering trauma also have changes to functional connectivity within the brain that can be mapped and quantified using MRI-driven techniques analyzing the default mode network (DMN). The DMN spans the hippocampus, precuneus, cingulum, inferior parietal lobe, and the orbital frontal, medial prefrontal, lateral temporal, lateral parietal, cingulate, and parahippocampal cortices ^{202 –204} and is activated during rest or self-reflective periods but suppressed during attention or goal-oriented tasks. ^202,205,206 The DMN is suspected to be involved in integrating multi-modal information and evaluating the cognitive and homeostatic environment, and is connected by white matter tracts across the brain. However, its exact function is still unknown. ²⁰³ The DMN is frequently used to characterize resting-state connectivity changes in neurological disorders, but has only recently been applied to examining functional outcomes of mTBI. ^205,207 A reduction in resting state functional connectivity (rs-FC) within the DMN during the acute stage of mTBI injury has been demonstrated, ²⁰² consistent with significant reductions in rs-FC in more severe TBI populations shortly after injury. ^208,209

Considerable variability exists in DMN rs-FC reductions in patients with mTBI during the subacute stage of injury across studies. These inconsistencies may be attributed to differing times of observation post-injury, heterogeneous inclusion criteria, and the innate differences in patients' rate of recovery. ²¹⁰ Patients with mTBI have rs-FC within the DMN at the chronic stage of injury. ^202,210,211 Interestingly, patients with mTBI at the chronic stage have increased rs-FC between the DMN and the task positive network (TPN). ²⁰² Additionally, there is increased rs-FC between the TPN and the medial prefrontal cortex (MPFC) and dorsal posterior cingulate cortex (dPCC) nodes of the DMN, with concomitant increases in rs-FC between the DMN and the left dorsolateral prefrontal cortex (L-DLPFC). ²⁰² An increase in rs-FC between the DMN and dorsal anterior cingulate cortex (ACC) and bilateral insula has been observed at both chronic ²⁰² and subacute stages. ²¹² The bilateral insula is part of the salience network (SN), ²¹³ and is sometimes co-activated with the TPN. ²¹⁴ This increased connectivity between the DMN and regions of the SN led to suggestions that there is a demand for the SN to modulate the activity of the disrupted DMN following mTBI, ²⁰² and that the increased modulation is a possible cause of increased cognitive fatigue seen in patients with mTBI. ²¹⁵ The changes in functional connectivity were found to persist for up to 6 months post-injury and likely reflect DAI in white matter. ²⁰²

Cerebral blood flow (CBF) in the brain is also altered upon mTBI. ²⁰² Imbalances between resting arterial spin labeling (ASL) perfusion within the DMN and TPN are found at the chronic stage, ²⁰² leading to suggestions that recovery from mTBI is a highly dynamic mechanism that involves CBF changes to balance out the basal activity of the two networks. ²⁰² The altered CBF allocation to the two networks may also contribute to the cognitive fatigue, ²¹⁵ and attentional deficits seen in patients with mTBI. ²¹⁶

Patients suffering from persistent post-concussive symptoms (PPCS) fail to maintain innate balance between perfusion to the DMN and TPN. ²⁰² An increase in resting perfusion in the TPN nodes is seen in patients suffering from PPCS, ²⁰² and patients with PPCS were unable to uphold network perfusion patterns comparably with non-PPCS mTBI and control participants. Therefore, it has been suggested that during the initial stages of injury, it may be possible to use network perfusion patterns to determine which patients will develop PPCS. ²⁰² However, the relevant data involved two groups of patients with mTBI of varying ages and changes in perfusion may be due to the age difference, ²⁰² further confounded by the fact that older patients are at greater risk for developing PPCS. ²¹⁷

Following mTBI, common complaints include cognitive fatigue, headaches, and emotional and vestibular dysregulation. ^11,12,218 Children may also experience mood changes, sleeping difficulties, and concentration and memory issues. ²¹⁹ Despite this, neuropsychological tests rarely detect differences in objective performance on working memory tasks in children after mTBI. ²¹⁸ Children and adolescents represent the population that is most at risk for incurring an mTBI, ²²⁰ hence the Australian Barlow group including the KidStim Laboratory, focus on mTBI in children. ²¹⁸ Working memory is described by the Barlow group as the ability to hold recent information and transiently manage it for goal-directed behaviors. ²¹⁸ Working memory-related cortical activation is decreased in children who have not recovered from mTBI at 1 month post-injury. ²¹⁸ Additionally, similar regions of cortical hypoactivation have been found in children 41 days post-injury. ²²¹ However, other studies have demonstrated hyperactivation in cortical areas. ^{222 –225} These inconsistent findings are possibly the result of heterogeneous memory tasks, inconsistent inclusion criteria, and differing age ranges between studies. ²¹⁸ Although the Barlow group found there to be similar working memory-related cortical activation and neurocognition in patients with mTBI and healthy controls, children showing signs of PPCS demonstrated greater inhibition of the DMN and hypoactivation in the DLPFC during the working memory task, compared with children showing no sign of PPCS post-injury. ²¹⁸ Despite these changes, no performance differences were detected. ²¹⁸ Resting-state data therefore provide an insight into internodal activity and a useful platform to characterize changes post-injury; however, more research is needed to achieve accurate prognostic value.

Metabolic changes

In addition to structural and functional aberrations in white matter following injury, a number of metabolic changes reflective of gliosis or pathology can be detected using magnetic resonance spectroscopy techniques. Magnetic resonance spectroscopy has been used to detect reductions in ratios of N-acetylaspartic acid (NAA), choline (Cho), and creatine (Cr) in the genu of the corpus callosum following mTBI in athletes regardless of number of hits. ³⁹ Athletes recovering from their first mTBI showed the greatest alteration in NAA/Cho and NAA/Cr ratios, perhaps indicative of a tolerance effect with further hits. ³⁹ Decreases in NAA/Cr ratios have been observed in the thalamus in mTBI patients with impaired cognitive function, and an increase in thalamic Cho/Cr ratio was evident in mTBI patients with self-reported sensory symptoms. ²²⁶ However, the NAA/Cho and NAA/Cr ratios do not differ with the severity of symptoms. ²²⁷ A substantial literature explores the relationship between metabolite changes in the brain following TBI, including in white matter. These have been extensively reviewed, ^228,229 and will not be covered further here.

Quantitative susceptibility mapping (QSM) is a novel imaging technique that allows in vivo measurement of magnetic biometals (iron) and high concentrations of Ca²⁺. ^230,231 The technique has displayed sensitivity in detecting changes in neurodegenerative disease, such as Parkinson's disease ²³⁰ and Alzheimer's disease, ²³² and has been used in TBI research. ^233,234 Iron deposition in the cortex has been associated with cognitive decline, and is believed to induce oxidative stress and a pro-inflammatory response. ²³² Abnormal accumulation of iron in white matter can be caused as a result of microhemorrhage following TBI, ²³³ and in patients with severe TBI, iron oxide targeting P-selectin microparticles can be used to identify endothelial activation in the cortex and hippocampus. ²³⁵ In a susceptibility-weighted imaging study in patients with chronic mTBI, abnormal iron depositions were detected in the hippocampus, thalamus, and the splenium of the corpus callosum at 6 months post-injury. Additionally, iron accumulations in the right substantia nigra are positively associated with cognitive impairments in patients with mTBI. ²³⁶

More recently, QSM has been utilized in a FPI rodent model to visualize thalamic Ca²⁺ influx in severe TBI, repeated mTBI, and single mTBI. No calcifications were detected 1 week post-injury; however, at 4 weeks calcifications were detected in all injury severities. The repeated mTBI group had the largest calcification volumes, a higher prevalence of calcifications compared with control and sham animals, and calcium increases in repeated hits were associated with neurological disability. ²³⁴ Another novel technique is flortaucipir positron emission tomography (PET), which allows the study of tau pathology in relation to TAI. Flortaucipir PET allows for in vivo detection of abnormal tau in neurofibrillary tangles, which traditionally has been used as a marker of pathology in post-mortem Alzheimer's disease tissue. In patients who received a single moderate TBI, flortaucipir binding was increased many years after injury compared with healthy controls. This increase in flortaucipir binding was associated with the presence of TAI, and with CSF markers of neurodegeneration T-tau and P-tau. ²³⁷ Future analysis of biometals and metabolites may provide insight into subtle changes in white matter in patients with TBI.

Glutamate excitotoxicity and changes in calcium ions and reactive species

In addition to the primary mechanical insult, white matter injury induces a cascade of molecular changes within neuroglia that can be detected using biochemical and imaging based methods. Glutamate-induced excitotoxicity is a prolonged or excessive exposure to glutamate that ultimately results in neuronal and glial cell death. ^238,239 In normal physiology, glutamate is rapidly removed from synapses by glutamic acid reuptake systems on glial cells and neurons. ²⁴⁰ Reuptake prevents a buildup of glutamate within synapses, sustaining glutamate levels at approximately 0.6 μmol/L. ²⁴¹ However, after neurotrauma the concentration of extracellular glutamate can increase to up to 5 μmol/L. ²⁴¹ Following TBI in humans, the levels of glutamate in the CSF are increased, and remain higher 1 week following injury, with these levels correlated with injury severity. ²⁴² Further, by 2 days following a moderate midline FPI, tonic glutamate levels are increased by 178% in the striatum and 256% in the dentate gyrus. ²⁴³

Glutamate excitotoxicity has also been implicated in white matter damage following SCI. ²⁴⁴ Glutamate elevation following injury is caused by: 1) Ca²⁺-dependent excitotoxic release of glutamate from axons; 2) dysfunctional astrocytic and neuronal glutamate reuptake and recycling; 3) ATP-dependent P2X₇receptor-mediated astrocytic glutamate release; 4) spillage from adjacent injured cells and; 5) glutamate release from activated microglia, associated with an inflammatory response. ^{241,245 –247} In addition, there is a positive feedback mechanism associated with glutamatergic excitotoxic insult. Glutamate interacts with N-methyl-D-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors to increase the expression of inflammatory cytokines and chemokines. ²⁴⁸ These toxins stimulate nearby microglia and astrocytes to release more glutamate and inflammatory factors into the extracellular space, exacerbating excitotoxic insult. ²⁴⁹ However, increased extracellular glutamate alone cannot induce the full spectrum of toxic effects observed following CNS injury. ²⁵⁰

Intracellular Ca²⁺ overload activates apoptotic cellular mechanisms and has been coined the “final common pathway of cell death.” ²⁵¹ After neurotrauma, there is an immediate rise in extracellular Ca²⁺ levels due to spillage from injured and inflamed cells, and release from intra-axonal Ca²⁺ stores. ²⁵² Using ⁴⁵ Ca autoradiography following a lateral FPI, there is a marked increase in ⁴⁵ Ca accumulation, even in areas of the brain without gross morphological damage, showing that Ca²⁺ flux plays a critical role in TBI pathology. ²⁵³ Ca²⁺ rapidly enters neuroglia through a variety of channels and receptors, including AMPA receptors, NMDA receptors, voltage-gated calcium channels (VGCCs), and P2X₇ receptors, ²⁵⁴ as well as through membrane pores created by the shearing force during the initial mechanical insult. ²⁵⁵

Increased intracellular Ca²⁺ is self-propagating, due to Ca²⁺-induced release of Ca²⁺ from intracellular stores, such as the endoplasmic reticulum, as well as reversal of the Na⁺/Ca²⁺ exchanger, which then pumps additional Ca²⁺ into the cell. ^{256 –258} Changes in Ca microdomains are also apparent in optic nerve vulnerable to secondary degeneration, detected using nanoscale secondary ion mass spectroscopy. ²⁵⁹ Following neurotrauma, oligodendrocytes and OPCs are particularly vulnerable to intracellular Ca²⁺ overload, ⁴⁰ partially because they contain higher concentrations of P2X₇receptors and AMPA receptors than neurons and other glial cells. ^260,261 Therefore, more receptors are activated after injury, leading to greater Ca²⁺ influx when compared with cells that have a lower concentration of these receptors. ²⁶² Further, oligodendrocytes and OPCs do not have the capacity for receptor-mediated desensitization of AMPA receptors when overstimulated, thus increasing their risk of Ca²⁺-mediated damage. ²⁶³ Once Ca²⁺ has entered cells, it mediates a variety of downstream pathways, including mitochondrial dysfunction and oxidative stress, ultimately resulting in apoptotic cell death and the spread of secondary degeneration, ²⁶⁴ with oligodendrocytes being particularly vulnerable.

Oligodendrocyte susceptibility to oxidative stress

Increased intracellular Ca²⁺ accumulates in mitochondria, resulting in an increase in mitochondrial and cytoplasmic production of reactive oxygen species (ROS) and reactive nitrogen species (RNS), ¹³⁰ as shown in Figure 1.

OPEN IN VIEWER

FIG. 1.

Schematic diagram showing intracellular production and effect of reactive oxygen species during secondary degeneration. Reactions shown are indicative of processes and are not balanced equations. Not all possible reactions are shown., ETC, electron transport chain; MPT, mitochondrial permeability transition; OMM/IMM, outer/inner mitochondrial membrane; VDAC, voltage dependent anion channel. Adapted fromO'Hare Doig et al. ²⁶⁵

Following intracellular Ca²⁺ influx, Ca²⁺ accumulates inside the mitochondria via the mitochondrial permeability transition. ²⁶⁶ Mitochondrial protein phosphatase then dephosphorylates cytochrome c oxidase, reducing allosteric ATP-inhibition and increasing the mitochondrial membrane potential. ²⁶⁷ As ATP synthesis increases through the Krebs cycle, electrons are leaked at complexes I and III of the electron transport chain on the inner mitochondrial membrane, resulting in the increased formation of the superoxide anion (O₂ ^•-). ^268,269 Superoxide dismutases reduce O₂ ^•- to hydrogen peroxide (H₂O₂), and the iron-catalyzed Fenton reaction further dismutases H₂O₂ into hydroxyl radicals (OH^•). ^270,271 OPC vulnerability is heightened by increased baseline intracellular iron levels, ²⁷² as increased intracellular iron is associated with increased ROS production via iron-facilitated Fenton reactions. ²⁷³

Our team has demonstrated that at 1 and 3 days post-injury, there is an increase in H₂O₂ following a partial optic nerve injury. ²⁶⁵ Nitric oxide (^•NO) is enzymatically synthesized by inducible nitric oxide synthase (iNOS), and reacts with O₂ ^•- to produce peroxynitrate (ONOO^-). ²⁷⁴ ONOO^- can then be further reduced to form OH^• and nitrogen dioxide (^•NO₂). ²⁷⁵ O2^•- can travel from the mitochondria into the cytoplasmic space through voltage-dependent anion channels on the outer mitochondrial membrane, resulting in elevated ROS outside of the mitochondria. ²⁷⁶ Further, H₂O₂ is membrane permeable, and so can diffuse out of the mitochondria into the cytosol. ²⁷⁷ Within the cytoplasm and nucleus, these ROS oxidize DNA and lipid structures, and nitrate proteins, causing cellular damage and apoptosis. ²⁶⁴ OH^• also deprotonates lipids and proteins, causing a self-propagating cycle of oxidative damage. ²⁷⁰

Our team has shown that following a partial optic nerve transection, the immunoreactivity of carboxymethyl-lysine (CML), an advanced glycation end-product as a marker of oxidative stress increases significantly post-injury in areas of nerve vulnerable to secondary degeneration, colocalized with Olig1+ oligodendroglial cells. ¹³⁰ When animals were treated with a combination of ion channel inhibitors aimed to reduce Ca²⁺ influx through AMPA receptors, P2X₇ receptors, and VGCCs, the immunointensity of CML decreased to normal levels, suggesting increased advanced glycation end-products in this model are at least in part a result of Ca²⁺-dependent mechanisms. ¹³⁰

In a normal, uninjured system, ROS and RNS are detoxified by antioxidants, such as glutathione peroxidase (GPx1), heme oxygenase-1 (HO-1), and manganese superoxide dismutase (MnSOD), preventing a buildup of free radicals within cells. ⁷⁸ Cells respond to increased production of ROS and RNS by increasing production of antioxidants; with research from our team showing a significant increase in MnSOD immunoreactivity in the optic nerve as early as 5 min after partial transection, colocalized within astrocytes. ^278,279 By 3 days following injury, there is a significant increase in GPx1, often colocalized within CC1+ oligodendrocytes, with a significant increase in HO-1 by 7 days post-injury, predominantly colocalized inionized calcium-binding adaptor molecule (Iba1) positive microglia. ²⁶⁵ Following two repeated closed-head weight-drop mTBIs in rats, MnSOD immunoreactivity increases 11 days post-injury. ²⁸⁰ When a combination of ion channel inhibitors to limit Ca²⁺ entry via AMPA receptors, P2X₇ receptors, and VGCC is used following repeated mTBI, MnSOD immunoreactivity decreased, indicating modulating Ca²⁺ influx is key to limiting this antioxidant response. ²⁸⁰ At both 1 and 14 days post-injury following a spinal cord contusion in rats, the level of MnSOD is significantly increased, with a similar increase in copper zinc superoxide dismutase (CuZnSOD) and GPx1 at 14 days post-injury. ²⁸¹

When the rate of ROS and RNS generation becomes greater than the rate at which antioxidants can counterbalance, cells are unable to prevent the damaging effects of these free radicals and undergo oxidative and nitrosative stress. ²⁴⁹ OPCs have lower concentrations of endogenous antioxidants, such as GPx1 and MnSOD and are thus particularly vulnerable. ^272,282 Glutathione prevents cell apoptosis by conjugating 4-hydroxynonenal (HNE), a by-product of lipid peroxidation, and suppressing the toxic activity of 12-lipoxygenase. ^283,284 Therefore, lower levels of glutathione are correlated with an increased risk of oxidative damage to lipids. ²⁸⁵ Further, when OPCs are engineered to overexpress the antioxidant MnSOD, there is an observable decrease in apoptosis. ²⁸⁶

Mitochondrial oxidative stress also induces endoplasmic reticulum (ER) stress. ²⁸⁷ ER stress results in unfolded proteins, ²⁸⁸ and if the stress persists for an extended period, it can result in cell death. ²⁸⁹ Further, in periods of stress, the ER releases large quantities of Ca²⁺, which in turn increase the production of ROS/RNS in the mitochondria. ²⁸⁷ Our research has shown that mitochondria in both axons and glia develop subtle structural changes following injury, with increased autophagic profiles demonstrating susceptibility of mitochondria to secondary degenerative processes. ²⁹⁰

Oligodendrocyte vulnerability to oxidative stress is further exacerbated by it role in maintaining the myelin sheath. ²⁸² The maintenance of myelin requires large amounts of energy, and involves catalyzing the production of lipids. ²⁹¹ One by-product of these reactions is ROS; therefore, there is an intracellular accumulation of ROS within oligodendrocytes, amplifying oxidative stress mechanisms. ²⁸²

Lipid peroxidation following injury

Lipid peroxidation, the process where ROS “steal” electrons from a cells membranous lipid bilayer, is an important downstream mechanism of oxidative stress. ²⁹² ROS attack the carbon to carbon double bonds of lipids, in particular of polyunsaturated fatty acids, as well as glycolipids, phospholipids, and to a lesser extent cholesterol. ²⁹³ This attack on lipids results in the formation of toxic aldehydes, such as HNE. ²⁹⁴ Further, iron plays a key role in ROS-mediated lipid peroxidation and enhances cytokine-mediated mitochondrial and membrane damage. ^81,295 Therefore, during secondary degeneration, OPCs not only produce more ROS but have greater ROS-mediated cell damage than other neuroglia due to their increased intracellular iron levels. ²⁷² HNE mediates cell damage by binding to and disrupting membrane transport systems, including impairing glutamate reuptake in astrocytes. ^296,297 Therefore, the production of HNE by lipid peroxidation causes a positive feedback for glutamate excitotoxicity and exacerbates the cycle of cell damage. ²⁹⁸

Our team has shown overall increased levels of HNE from 3 days after partial injury to the optic nerve, ²⁶⁵ with a significant increase in HNE within the OPC and mature oligodendrocyte populations at this time. ⁴³ When a combination of three ion channel inhibitors targeting AMPA receptors, P2X₇ receptors, and VGCCs was delivered acutely following the injury, OPCs were preserved when HNE decreased. The protection was associated with improvements in visual function, suggesting that limiting lipid peroxidation may aid in maintaining the number of OPCs following injury. ⁴² Following a clip-compression-induced SCI, there is a biphasic increase in lipid peroxidation as shown by malondialdehyde assays, with the first peak at 4 h post-injury, and the second increase at 24 h to 5 days post-injury. ²⁹⁹ In a model of repeated mTBI, our team has also shown increased lipid peroxidation in cortical neurons, demonstrating that oligodendroglia are not the only cell type vulnerable to lipid peroxidation. ³⁰⁰ Further, following TBI in humans, circulating plasma levels of HNE increase compared with control patients. ³⁰¹

Protein nitration following injury

Nitrosative stress also has a neurodestructive effect, mediating further lipid peroxidation, as well as protein nitration. ^302,303 Protein nitration occurs when a nitro group is added into the structure of a protein, ³⁰⁴ and primarily occurs as the reaction between ^•NO or ^•NO₂ and the amino acid tyrosine, to form 3-nitrotyrosine (3-NT). ³⁰⁵ Immunointensity of 3-NT, a protein nitration indicator, is significantly increased in a diffuse distribution 3 days following partial injury to the optic nerve. ⁴² Further analysis using NanoSIMS technology shows that 3-NT is increased at 3 days within mature oligodendrocytes, NG2+ cells, Olig2+ cells, βIII tubulin+ axons, and paranodes, with no structure selectively vulnerable to damage. ⁴³ In addition, following a moderate diffuse closed-head injury, 3-NT immunoreactivity increased at 48 h post-injury. ³⁰⁶ This suggests that the nitrosative pathways also contribute to the cellular damage observed in secondary degeneration.

Oxidative DNA damage following injury

DNA is highly vulnerable to oxidative damage, and there are a variety of mechanisms through which free radicals mediate DNA damage during oxidative stress, one of which is nucleobase modification. ²⁴⁹ These DNA lesions can have extremely detrimental effects, by either altering the genetic code or by acting as a block during DNA replication. ³⁰⁷ DNA is highly vulnerable to oxidative damage, as ROS, particularly •OH, can mediate hydroxylation of DNA nucleic acid bases. ²⁴⁹ This process converts deoxyguanosine to 8-hydroxy-2-deoxyguanosine (8OHDG), through a spontaneous transversion mutation, altering DNA structure. ³⁰⁸ Our team has shown that by 24 h following partial optic nerve injury, 8OHDG is significantly increased in the ventral optic nerve white matter, and this is associated with apoptotic cell death. ^265,309 However, in our closed-head weight-drop model of repeated mTBI, there is no increase in 8OHDG immunoreactivity in a range of brain regions, with either 1, 2, or 3 injuries at 3 months following injury, ³¹⁰ or with two injuries at 11 days post-injury. ²⁸⁰ More work is needed to assess the acute time course of 8OHDG immunointensity in this model.

Going on to define the cellular specificity of the changes, we have shown that oligodendroglial cells are significantly more likely to succumb to oxidative 8OHDG DNA damage following neurotrauma compared with other neuroglial cells and structures. ⁴³ Three days following the optic nerve injury, both OPCs and oligodendrocytes had increased levels of 8OHDG DNA damage. ⁴³ Newly derived oligodendrocytes exhibited significantly more DNA damage than pre-existing oligodendrocytes, suggesting newly differentiated cells are more vulnerable to oxidative damage. ⁴³ The DNA-damaged oligodendrocytes were more likely than non-DNA-damaged oligodendrocytes to become apoptotic. ⁴³ However, newly derived oligodendrocytes were less likely to be apoptotic than pre-existing oligodendrocytes. ⁴³ Thus, despite their increased susceptibility to DNA damage, newly derived oligodendrocytes are less vulnerable to apoptotic cell death than their pre-existing oligodendrocyte counterparts. ⁴³ Interestingly, by 28 days after injury, these newly derived oligodendrocytes are still less likely to be apoptotic, but they also have decreased expression of MyRF. ⁴³ This suggests that newly derived DNA damaged oligodendrocytes are more likely to survive following injury but have a reduced myelination capacity. ⁴³ Therefore the ability for remyelination is compromised following oxidative damage due to neurotrauma and may lead to poor repair of damaged myelin. ⁴³

Changes in numbers of oligodendroglia following injury

It has been widely reported, including in Australian research, that following experimentally induced TBI and SCI there is a loss of oligodendrocytes. ^{40,41,311 –314} However, some studies have not detected oligodendrocyte death, ¹²⁰ perhaps indicating differential responses of these cells depending on injury severity or mode.

We have shown that following partial injury to the optic nerve white matter, ⁴³ using Olig2 together with NG2 as markers for OPCs, the number of OPCs remains similar to control at 1 and 3 days after injury. ³¹⁵ However, OPC densities significantly decline from 7 days to 3 months, which is not attributable to optic nerve swelling. ³¹⁵ At 3 days post-injury, OPCs proliferate more than controls. ³¹⁵ At 1 and 3 days post-injury, the density of CC1+/Olig2+ immature oligodendrocytes decreases; however, by3 days there is an increase in the density of CC1+/Olig2- mature oligodendrocytes, indicating differentiation of a subset of surviving immature oligodendrocytes. ³¹⁵ By labeling dying oligodendroglia with the cell death marker TUNEL, it was found that an increase in TUNEL+ staining occurred across the oligodendroglial lineage by 7 days following injury, with a significant increase in OPC cell death sustained at 1 month post-injury. ³¹⁵ This suggests that despite acute proliferation, there is an overall chronic depletion in the OPC population. ^43,315

Similarly, following a contusion SCI, numbers of BrdU+/NG2+ OPCs increase, with proliferation of OPCs tripling in the first week following injury. ³¹³ However, within the first 7 days following injury, the total number of OPCs decreases, demonstrating an overall reduction of this cellular population despite the proliferative response. ³¹³

Change in the OPC population may be slightly different following TBI. Using an FPI model, the number of OPCs increases in white matter tracts from 2 days to 21 days following injury, ⁴¹ quantified based upon use of Olig2 or Tcf4 individually to indicate OPCs. Further, although BrdU+/PLP+ newly derived oligodendrocytes are rare in the corpus callosum, they are significantly higher in number following a CCI than sham injury. ¹²⁰ Finally, in another study utilizing a CCI, immunoreactivity of the OPC marker NG2, quantified by optical density measurements, peaks at 1 to 4 days following injury in the corpus callosum. ³¹⁶ Taken together these studies suggest that OPC populations increase in response to injury in white matter following TBI. In the first few days following a moderate to severe TBI in humans, oligodendrocytes are lost with a concomitant increase in the number of OPCs. ³¹⁷ Despite these findings, it is possible that variability exists in number of oligodendroglia depending on the type of white matter injury.

Axonal loss and neuronal damage following injury

Neuronal loss both within and around the injury site is a hallmark of white matter injury. ³¹⁸ In TBI, neural degeneration has been observed within minutes following injury, ³¹⁹ featuring the degenerative cascade of TAI. ³³ Certain areas of white matter are more vulnerable to TAI, such as the corpus callosum and the brainstem. ^171,320 Staal and Vickers found that in vitro unmyelinated axons are more vulnerable to stretch injury that myelinated axons. ³²¹ Further, unmyelinated fibers in the corpus callosum are more selectively vulnerable to damage than myelinated axons in vivo. ³²² Following injury, increased intracellular Ca²⁺ within neurons results in damage to the axonal cytoskeleton and impairments to axonal transport via activation of degenerative enzymes and cysteine proteases such as caspase-3 and the Ca²⁺-dependent neutral protease calpain. ³²³ Caspase-3 cleaves proteins associated with DNA repair, contributing to the DNA damage observed following injury. ³²⁴ The damage cascade results in accumulation of axonal transport proteins within the axon, DNA fragmentation, and axonal swelling, with affected neurons eventually undergoing either apoptotic or necrotic cell death, impairing function. ³²⁵ In addition, DNA damage from ROS enhances post-traumatic neuronal death following injury. ³²⁶

The partial optic nerve transection model has allowed specific insight into the changes to axons and their parent somata vulnerable to secondary degeneration in a white matter tract. We have shown that at 24 h after injury, retinal ganglion cell somata show increased c-jun expression, associated with pro-apoptotic mechanisms. ²⁷⁸ By 3 days after injury, poly (ADP-ribose) polymerase (PARP) is upregulated in axons in the ventral optic nerve, implicating necrosis as an additional pathway to cellular demise. ³²⁷ There is also decreased axonal stability at 1 day following PT, shown through decreased acetylated tubulin expression. ⁴² Further, there are no changes in the ratio of Ca²⁺-independent phosphorylated Tau/panTau at [S ²⁶² ], whereas there is increased Ca²⁺-dependent phosphorylated Tau isoforms at this time. ⁴²

When treated with a combination of ion channel inhibitors that reduce Ca²⁺ influx through AMPA receptors, P2X₇ receptors, and VGCCs, the ratio of phosphorylation of both [S ³⁹⁶ ] and [T ²⁰⁵ ] relative to total Tau decreases significantly at 3 day post-injury compared with injured untreated controls. ⁴² The ion channel inhibitor combination also ameliorates acetylated tubulin immunoreactivity. ⁴² Conversely, NogoA immunoreactivity decreases in the ventral optic nerve with injury but increases when animals were treated with the ion channel inhibitor combination. However, changes to NogoA did not appear to be a driver of therapeutic benefit following ion channel inhibitor administration, and the functional significance of changes to NogoA remain unclear. At 3 months following the optic nerve injury, β-III tubulin immunointensity is decreased, indicative of loss of axonal integrity and axonal death. ¹³⁰ Indeed, by 3 months after injury, the number of axons in the ventral optic nerve has decreased by 30.7%. ³⁴ The ion channel inhibitor combination protects β-III tubulin immunoreactivity 3 months post-injury, associated with preservation of myelin and improvements in visual function. ¹³⁰ There is also a significant decrease in retinal ganglion cell densities in the central retina both 1 and 3 months after these injuries, which suggests retrograde damage to cell bodies along axons. ³²⁸ Taken together the data indicate that loss of neurons is associated with Ca²⁺ dysregulation, oxidative stress mechanisms, and dysmyelination, and likely contributes to a decrease in function. ^42,130

These degenerative changes to neurons and their axons are not limited to the optic nerve injury model. Following mTBI with TAI, degenerating axons are apparent with electron microscopy at 3 days through to 6 weeks post-injury, with swollen mitochondria, either abnormally high or low cytoplasmic density, and vesicle accumulation. ¹²⁰ Unmyelinated axon density decreases in the corpus callosum following a moderate FPI, with diameter decreasing from 3 to 7 days post-injury and recovering at 15 days. ³²⁹ In a TAI study utilizing compound action potentials evoked in the corpus callosum, it was found that the amplitude of myelinated axons was decreased up to 3 days post-FPI but recovered to control levels by 7 days. Whereas the unmyelinated axonal amplitude did not recover, therefore a differential vulnerability was observed. ³²² Following a single-impact open-head weight-drop TBI, axons are damaged, with a concomitant decrease in conduction velocities across the corpus callosum. ³³⁰ However, there is no increase in TUNEL+ neurons in white matter following a FPI, which suggests neurons are less vulnerable to apoptotic cell death than oligodendroglia. ³³¹

Tau hyperphosphorylation is a driver of neurodegeneration as it results in a disruption of microtubules and may result in interrupted axonal transport and therefore decreased synaptic transmission. ³³² The Corrigan team at the University of Adelaide (Collins-Praino and colleagues ³³³ ) have shown that regardless of whether rats receive a single moderate to severe TBI or three mTBIs, the level of AT180 tau phosphorylation has a biphasic pattern, with an initial spike at 24 h following injury, which returns to baseline before increasing again at 3 months post-injury. Further, levels of protein phosphatase 2Ac (PP2Ac) decrease at both 24 h and 3 months following injury, hypothesized to demonstrate a loss of PP2A phosphatase activity. ³³³ Overall, this indicates that tauopathy and tau phosphorylation may be a significant feature of neurodegeneration following both repeated mTBIs and more moderate to severe single TBIs.

Axonal loss is also a hallmark of SCI. By 1 week following a contusion SCI, only 13% of axons remained in the degenerated core of the dorsal column at the injury site. ³³⁴ Axons swelled, with the axoplasm becoming dense, followed by myelin detaching from the axon. ³³⁴ Axons subsequently degenerated, leaving a microcyst inside a slowly disappearing myelin sheath. ³³⁴

The amyloid-β (Aβ) peptide is derived from amyloid precursor protein (APP), with the aggregation of Aβ believed to play a role in Alzheimer's disease pathogenesis. ³³⁵ Aggregation of Aβ is evident following severe TBI in humans, ³³⁶ so it is not surprising that there is an epidemiological association between TBI and Alzheimer's disease later in life. ⁶ The Vickers team have recently shown (Collins and associates ³³⁷ ) that 30 days following a lateral FPI in the APP/PS1 transgenic model of Alzheimer's disease in mice, the aggregation response of Aβ is dependent on the state of amyloidosis at time of injury, with pre-plaque mice showing increased Aβ deposition in the cortex, and mice with established amyloidosis showing decreased total Aβ deposition. More work is needed to determine the Aβ response in white matter following injury. The functions of APP are manifold, including roles in neuronal and synaptic processes. ³³⁸ The Vink team has shown (McAteer and co-workers ³³⁹ ) that following repeated mTBI, there are increases in APP immunoreactivity at 12 weeks post-injury. However, APP derivatives can mediate both neurotoxic and neuroprotective effects. Indeed, the team in South Australia (Corrigan and colleagues ³⁴⁰ ) has gone on to show that if APP is knocked out in mice, there is increased cortical and hippocampal cell loss, axonal injury, and decreased behavioral deficits following diffuse mTBI, suggesting upregulation of APP plays a neuroprotective role. ³⁴⁰ Exogenous treatment with the APP derivative sAPPα prevents deficits in APP knockout mice after injury. ³⁴¹ Together with Roberto Cappai, the team have shown (Plummer and associates ³⁴² ) that this neuroprotective effect of APP is mediated by residues 96-110 in the heparin-binding site in the growth-factor like domain (D1) of sAPPα, with APP96-110 having a neuroprotective effect following focal CCI. Intravenous administration of APP96-110 30 min after a moderate to severe diffuse impact-acceleration TBI results in reduced motor deficits, and decreased axonal injury and neuroinflammation in the corpus callosum at 3 days following injury. ³⁴³ Further, when APP96-110 is administered intravenously 5 h post-injury, there are improvements in motor deficits and axonal injury in the corpus callosum. ³⁴³ Therefore, APP96-110 may be a potential therapeutic option to limit axonal injury in white matter following TBI.

Dysmyelination and myelin structural changes

The myelin sheath is predominantly lipid-based, which means it is highly vulnerable to oxidative stress and lipid peroxidation, ³⁴⁴ resulting in widespread dysmyelination following white matter injury. ^15,345,346 Dysmyelination is a destructive process whereby the myelin sheath becomes abnormal in structure, associated with a loss of function due to reduced conduction of neural signals. ³⁴⁷ A hallmark feature of dysmyelination is myelin decompaction. Myelin decompaction occurs when the myelin lamellae separate and the paranodal loops detach and split, associated with increased intracellular Ca²⁺, as shown in Figure 2. ³⁴⁸

OPEN IN VIEWER

FIG. 2.

Characteristics of dysmyelination and myelin decompaction. Following CNS injury, myelin structures succumb to oxidative stress and lipid peroxidation. This results in paranodal splitting, myelin retraction and decompaction, Ca²⁺ influx, and cytoskeletal alterations to the nodal complex. Diffusion of nodal components results in impaired saltatory conduction along nerves, resulting in axonal injury and/or loss. As irregular myelin repair is attempted by newly myelinating oligodendrocytes, the number of atypical nodal complexes increases. Red circles represent nodal and paranodal components. CNS, central nervous system. Figure created with BioRender.com; modified from Podbielska et al. ³⁵⁰

Oxidative stress disrupts both the integral myelin sheath proteins, and the proteins between the surfaces of the myelin lamellae. ^351,352 We have shown that from 5 min following partial optic nerve injury, generalized increases in MBP immunofluorescence are observed in the ventral optic nerve, indicative of myelin degradation and increased immunoreactivity of MBP. ²⁷⁸ However, by 3 months following injury, MBP immunoreactivity is decreased, suggesting long-term myelin loss. ³⁴ In a rat model of TBI, MBP becomes proteolyzed in a calpain-mediated manner, contributing to lost integrity of the myelin sheath and dysmyelination. ³⁵³ MBP has also been found to be elevated in the CSF following severe TBI in humans. ³⁵⁴

Luxol fast blue (LFB) staining is a frequently used indicator of myelin integrity. Following a CCI, LFB staining reveals lesions within the corpus callosum from 1 day to 3 months after injury. ³⁵⁵ At 1 year following a parasagittal FPI in rats, abnormal LFB staining is observed in the corpus callosum, with a decrease in LFB staining in the ipsilateral cerebral peduncle compared with sham animals, indicating loss of myelin. ³⁵⁶ At 24 h and 2 days following a contusion SCI, LFB staining is considerably decreased in the dorsal column at the middle of the injury site, with more granular staining visible in the ventrolateral white matter; by 10 weeks post-injury, a thick rim of LFB staining is apparent surrounding the injury site. ³³⁴

Given the optic nerve is a highly myelinated white matter tract, the partial transection model has proved useful for studying changes to myelin structure following injury. We have used electron microscopy to demonstrate a significant increase in the proportion of axons with decompacted myelin in the ventral optic nerve at 3 months after injury, ³⁵⁷ associated with lipid peroxidation and oxidative stress as described above. ³⁵⁸ At 6 months, loosening of the myelin sheath is widespread throughout the nerve, with myelin thickness significantly increased at this time-point, likely due to the loosening. ³⁴ The number of intraperiodic lines surrounding myelinated axons also increases at 6 months post-injury, suggesting oligodendrocytes may be wrapping more layers of myelin around axons in an attempt at repair. ³⁴

In SCI, the necrotic core of the injury is surrounded by a rim of demyelinated axons. ³⁵⁹ After a contusive SCI, the number of demyelinated axons peaks at 1 day following injury, and starts to resolve by 7–14 days, before then increasing again up to 450 days post-injury. ³⁶⁰ Following a chronic clip compression injury of the rat thoracic spinal cord, the myelin is thinner in the injured dorsal column, associated with reduced action potential amplitudes in affected axons. ³⁶¹

Following mTBI with TAI, electron microscopy has revealed an increase in double-layered myelin, ¹²⁰ which can occur as myelin degenerates around injured axons. ³⁶² We have shown that in normally myelinated axons myelin thickness in the corpus callosum is not significantly altered with up to three repeated mTBIs. ³¹⁰ However, more work is needed to quantify dysmyelination, myelin decompaction, and loosening at various time-points following TBI at the electron microscopy level.

Dysmyelination severely disrupts the myelin node/paranode complex. The nodal sodium channels and the paranodal protein, Caspr, diffuse along the axon, increasing the size of the node and paranode respectively. ³⁶³ Our team has shown that at 1 day following partial optic nerve injury, there is a significant increase in both the paranodal gap length and the proportion of atypical nodal complexes, and these changes are maintained at 3 months after injury. ³⁶⁴ After closed-skull impact, paranodes were either shortened or absent with a concomitant increase in heminodes. ³³⁰ Following closed-head weight-drop mTBI, the length of the paranode, the node of Ranvier, and the proportion of atypical nodal complexes in the corpus callosum increases. ²⁸⁰ Interestingly, inhibiting VGCCs, P2X₇ receptors, and AMPA receptors results in improvements in both the length of the node as well as the proportion of atypical nodal complexes in both the optic nerve and mTBI models, indicating increased intracellular Ca²⁺ is a key mediator of myelin integrity at the node of Ranvier. ^42,280,365

Dysmyelination after injury is associated with higher axonal susceptibility to excitotoxic insult, ³⁶⁶ evidenced by a significant decrease in axonal density associated with dysmyelination both 1 and 3 months after partial optic nerve injury. ³⁵⁷ It is therefore not surprising that dysmyelination is associated with a loss of visual tracking behavior after partial optic nerve injury at both acute and chronic timepoints. ^42,130 Abnormal myelination at a single internode can be sufficient to block neural signal transduction for an entire axon. ³⁴⁴ White matter damage has also been associated with functional deficits following experimental models of TBI, with deficits dependent on the location and severity of the injury. ^367,368 Therefore, limiting the extent of myelin damage, both by attenuating dysmyelination and promoting successful remyelination following injury, may result in improved functional outcomes.

Inflammation and gliosis

Myelin degradation is both a cause and consequence of inflammation, as shown in animal models of multiple sclerosis where myelin debris regulates microglial activation. ³⁶⁹ Along with infiltrating macrophages, activated microglia secrete numerous cytokines, chemokines, ROS, and growth factors, affecting nearby cells. ³⁷⁰ Proinflammatory chemokines further recruit immune cells to the injury site, exacerbating damage, ^371,372 with depletion of microglia following a moderate FPI resulting in a therapeutic effect on neuronal loss early post-injury. ³⁷³ However, microglia have pleiotropic responses to injury, and can also mediate anti-inflammatory responses depending on environmental cues and exposure to specific cytokines. ³⁷⁴ In SCI, it has recently been discovered that microglia form a “microglial scar,” without which functional outcomes after injury are worsened. ³⁷⁵ From 3 days after partial transection to the optic nerve, we have shown that Iba1+ microglia and ED1+ macrophages become activated and their numbers increase in ventral nerve vulnerable to secondary degeneration and remotely in the brain. ^278,376

The Corrigan team have shown (Collins-Praino and associates ³⁷⁷ ) that following a single diffuse impact-acceleration mTBI, there is an acute increase in Iba1+ cells in the hippocampus at 6 days, which resolves chronically by 3 months post-injury. The finding built upon earlier demonstrations of a significant increase in cortical microglial activation at 24 h following three impact acceleration diffuse mTBIs, and a significant increase following both one and three injuries at 12 weeks. ³³⁹ Meanwhile, we have shown in a closed-head weight-drop model of repeated mTBI that microglia in the corpus callosum remain hypertrophic after two injuries at 3 months, ³¹⁰ but only after one injury in cortex, suggesting repeated mTBI may cause more chronic microglial activation in white matter tracts. From 3 days following mTBI with TAI, there is an increase in microglial activation in the corpus callosum, persisting through to 6 weeks post-injury. ^120,378

Further, following a closed-skull impact TBI with TAI, astrogliosis also significantly increases up to 6 weeks post-injury in the corpus callosum where axons are damaged. ¹²⁰ To form a barrier around the injury site, reactive astrocytes become hypertrophic, their processes become interlaced, and these cells show an increased expression of glial fibrillary acidic protein (GFAP). ³⁷⁹ We have shown that by 24 h after partial optic nerve transection, astrocytes display a hypertrophic morphology, with a significant increase in both the width of astrocytic processes and number of minor astrocytic processes. ²⁷⁸ Neurotrauma can activate at least two different types of astrocyte reactivity, A1 and A2, ³⁸⁰ nomenclature that mimics the M1/M2 macrophage/microglia terminology. ³⁸¹ However, as with macrophages and microglia, ³⁸¹ it is likely that astrocyte reactivity is a spectrum of various phenotypes, rather than distinct reactive states. ³⁸⁰ In particular, A1 astrocytes have been associated with neuroinflammation, whereas A2 astrocytes have been associated with the production of reparative trophic factors. ³⁸⁰

Following injury, it appears that there may be a mixture of these phenotypes present. Ablation of proliferating reactive astrocytes following a moderate TBI resulted in inflammation and neural degeneration, ³⁸² suggesting the importance of astrocytes in the reparative response. Further, following white matter ischemic injury, astrocytes promote OPC differentiation and oligodendrocyte generation by secreting BDNF, a mechanism that may occur within traumatic injuries. ³⁸³ However, reactive astrocytes also release inhibitory growth factors, cytokines, ROS, and glutamate that exacerbate injury and may have a detrimental effect on overall function. ^{384 –387} Recent data have shown that within the cortex, the astrocytic marker GFAP and A1 astrocyte phenotype marker C3 are upregulated from 8 h following both rat FPI and mouse CCI, with the microglial marker Iba1 increased at 24 h following injury. A peak was reached in both models 72 h after injury, suggesting astrocytes may become reactive prior to microglial activation. ³⁸⁸ However, it is important to note that not all GFAP+ astrocytes were C3+, suggesting there may be a mix of reactive astrocyte phenotypes present following injury. ³⁸⁸

These mechanisms may also be at play within the white matter following injury. In LFP TBI, reactive astrocytes are found 1 day post-injury in the subcortical white matter tracts, continuing at least until 1 month post-injury. ³⁸⁹ .At both 24 h and 12 months following a repeated closed-head injury, astrogliosis increases, but when only a single injury is administered, there is no increase in astrocyte reactivity at 24 h and only a mild elevation at 12 months. ¹²² However, at 24 months post-injury, a single mTBI produced mild astrogliosis, which was more severe following repeated injuries. ¹²¹ However, in our closed-head weight-drop model of repeated mTBI in rat, we find no astrogliosis with either one, two, or three injuries 3 months post-injury in the corpus callosum. ³¹⁰ However, more research is needed to determine whether microglia and astrocytes have beneficial or detrimental effects following CNS injury.