Epigenetics of Neural Repair Following Spinal Cord Injury

Abstract

Spinal cord injury results from an insult inflicted on the spinal cord that usually encompasses its 4 major functions (motor, sensory, autonomic, and reflex). The type of deficits resulting from spinal cord injury arise from primary insult, but their long-term severity is due to a multitude of pathophysiological processes during the secondary phase of injury. The failure of the mammalian spinal cord to regenerate and repair is often attributed to the very feature that makes the central nervous system special—it becomes so highly specialized to perform higher functions that it cannot effectively reactivate developmental programs to re-build novel circuitry to restore function after injury. Added to this is an extensive gliotic and immune response that is essential for clearance of cellular debris, but also lays down many obstacles that are detrimental to regeneration. Here, we discuss how the mature chromatin state of different central nervous system cells (neural, glial, and immune) may contribute to secondary pathophysiology, and how restoring silenced developmental gene expression by altering histone acetylation could stall secondary damage and contribute to novel approaches to stimulate endogenous repair.

Electronic supplementary material

The online version of this article (doi:10.1007/s13311-013-0228-z) contains supplementary material, which is available to authorized users.

Physical, Medical, and Cellular Sequelae After Spinal Cord Injury

Spinal cord injury (SCI) results from a physical insult that encompasses—either completely or incompletely—4 major functions (motor, sensory, autonomic, and reflex). The complex variety of motor, sensory, and autonomic problems that arise from SCI are a direct result of the level of their lesion, and the extent of spinal cord disruption following the development of secondary pathology. The most common injury mechanisms are 1) contusion of the spinal cord by external force injury [1]; 2) impact alone with more transient compression; 3) laceration and transection (most common in thin, thoracic segments); or 4) distraction—forcible stretching of the spinal column in the axial plane (reviewed in [2] and [3]).

The pathophysiological processes that lay the groundwork for the long-term deficits due to SCI comprise both primary and secondary phases of injury [4, 5]. Primary injury produces direct immediate, mechanical disruption of spinal cord function, but the more widespread mechanisms of secondary damage ultimately determine the final chronic extent of neurological deficits (summarized in Fig. 1). Even if axons were made capable of regenerating along multiple tracts after SCI, they are faced with a formidable barrier of secondary glial and inflammatory pathophysiology [6, 7]. Because of this, a clear understanding of pathophysiological mechanisms that produce primary and secondary damage after SCI is critical in order to facilitate targeted therapeutic intervention aimed at ameliorating the specific pathology of an individual injury [8].

Fig. 1.

Pathophysiological progression of spinal cord (SC) injury from (a) acute to (b) subacute/chronic stage involves shifts in state and distribution of i) resident microglia and invading inflammatory cells (neutrophils, monocyte-derived macrophages, and T-cells); ii) macrophages which can be polarized to either pro-inflammatory (M1) or anti-inflammatory (M2) state; iii) astrocytes, which first need to contain the lesion, but later interact with glial progenitors, fibroblasts, meningeal cells, microglia, and macrophages to form the physical and molecular barrier of the glial scar; iv) oligodendrocytes, which first undergo apoptosis when they dislodge from damaged axons, and later need to recreate myelin on spared axons; and v) many ascending and descending axons that are broken and need to either regenerate over long distances or sprout to form alternative circuitry

Secondary Damage Produces Multiple Therapeutic Challenges at Different Stages of the Repair Process

The immediate damage following SCI is to neurons located within, and passing through, the injury site that become physically damaged and sever axons from their targets. The spinal cord rapidly develops hemorrhage at the site of impact, disrupting local blood flow and producing a microenvironment of hypoxia and ischemia, which creates a further block to neural circuit function [9–11]. Spinal gray matter is believed to be irreversibly damaged within the first 1–3 h after injury, with white matter damage more extensive by 72 h after injury [12]. These time windows vary depending on the extent and level of lesion, and can be influenced by age and health status at injury. Given that early intervention that minimizes secondary damage may be the best way to enhance repair, it is important to establish how and when damage is propagated mechanistically, and understand how this varies between individuals. This is perhaps even more important in light of the fact that both primary and secondary pathophysiology in SCI shares common cellular and molecular mechanisms with other types of central nervous system (CNS) injury, such as traumatic brain injury (TBI), cerebral ischemia, and subarachnoid hemorrhage [13–15].

The mechanical injury site is not only the nidus of primary clinical symptoms, but also the core from which the post-injury inflammatory response orchestrates extensive secondary damage after SCI via a series of complex cellular and molecular interactions [16] (Fig. 1). In addition to the immediate activation of local microglia in the injury core, the blood–spinal–cord barrier disruption from hemorrhage and local inflammation [17] causes peripheral inflammatory cells to invade the lesion site with increased production of chemokines and cytokines of the IL-1 family [18–20]. Although initial immune responses are essential for clearing primary lesion damage, the recruitment and persistent activation of both CNS resident inflammatory microglia and peripheral infiltrates (including macrophages, monocytes, T lymphocytes, and neutrophils) increases the severity of secondary damage following SCI [21] (reviewed in [3]).

The secondary phase of SCI has 3 distinct phases—acute, subacute, and chronic—which each carry a different sequence of secondary damage cascades [1, 7, 22]. Progression from one stage to the next is driven by the shifting activation state of cells in and around the injury site, with each phase yielding distinct biochemical and cellular signatures (summarized in Fig. 1) [23]. Although this cascade of secondary damage clearly expands the injury size and increases cell and axonal loss, individuals can have broadly different primary and secondary inflammatory responses to very similar lesions, with vastly different clinical outcomes [19]. Similar variations are seen in multiple rodent models where the same lesion tested across different laboratories—even in genetically homogenous strains of rats or mice—can produce different functional outcomes, suggesting an epigenetic–immune contribution may underlie the extent of secondary damage. Some of these variations could also be attributed to the health or nutritional status of individuals at the time of, or following, injury, and also to their health in the postoperative care environment [24].

Given emerging evidence of the complex interplay between immune system activation, nutritional well-being and epigenetic variation, here we consider if the intrinsic epigenetic status of cells participating in SCI repair may contribute to the variation seen in the inflammatory response. If epigenetic mechanisms do regulate the plasticity and activation state of multiple cells involved in SCI pathophysiology, understanding their role in the development of individual lineages and using this to guide the critical timing of plasticity-promoting approaches will be essential to temporally target epigenetic plasticity after SCI. Here, we review whether enhancing the epigenetic plasticity of key cells participating in repair at different stages of SCI recovery might allow us to harness their developmental capacity to drive repair after injury. Key to this is understanding how different chromatin remodeling factors work together to ensure successful development.

Epigenetic Regulation Mechanisms Important for Development

All cells in a multicellular organism have an essentially identical genotype, and yet produce a wide range of differentiated cell types with very different gene expression profiles that drive divergent functions. This progressive restriction in gene expression in development results from a combination of state-dependent (stage- and lineage-specific) transcriptional activation that interact with epigenetic gene silencing mechanisms to restrict gene expression (reviewed in [25]). Epigenetic mechanisms thus provide the bridge that connects how the environment affects how each cell uses its genome [26–28]. Understanding mechanistic modifications that have occurred at the environment–genome interface during life could hold critical clues as to why different individuals display experience-dependent variations in learning, personality disorders, plasticity, and long-term responses to neurological disease or injury.

Chromatin Structure and Gene Expression

Chromatin has a highly malleable structure that provides the substrate for transcriptional processes that establish distinct cellular identities and states. Although the term “epigenetic” now also encompasses the activities of a myriad of noncoding RNAs for the purposes of this review we will focus largely on histone–DNA interactions [29]. The majority of covalent histone modifications occur at their N- terminals (which are highly conserved), which act as substrates for several different types of post-translational modifications that alter chromatin structure, including acetylation, methylation, adenosine diphosphate ribosylation, ubiquitylation, and phosphorylation. These modifications can be correlated with various nuclear functions, including replication, chromatin assembly, and transcription, as well as gene silencing. Distinct patterns of histone modifications act in concert with DNA methylation, noncoding RNAs, and transcription factors to generate “histone–epigenetic codes” that are read by effector proteins to modulate chromatin function and transcriptional output [30–34].

DNA Methylation

The covalent addition of a methyl group at the pyrimidine ring of cytosine residues is catalyzed by DNA methyltransferases (DNMTs) [35], and occurs in vertebrates almost exclusively at cytosine–phosphate–guanine dinucleotides [36–38]. DNA methylation is the primary impetus for recruitment of histone deacetylases (HDACs), and is essential for mammalian development, evidenced by the ultimate developmental lethality of DNMT knockout mice [39, 40]. A methylated cytosine can repress transcription either directly by precluding transcriptional activators or indirectly by promoting recruitment of inhibitory regulatory proteins [37]. The indirect mode of repression requires the specific binding of methyl–cytosine–phosphate–guanine–binding domain proteins to methylated DNA, mediating transcription normally through the recruitment of repressor complexes, usually containing HDACs (reviewed in [37, 41]). How DNA methylation contributes to the stability of gene expression state is reviewed elsewhere [33].

Histone Acetylation and Deacetylation

Acetylation at several different lysine residues in the N-terminal tails of histone proteins generally delineates zones of open chromatin and gene activation. Acetylation opposes and neutralizes the positive charge of histone proteins, decreasing the attraction between histones and negatively-charged DNA [29]. This relaxed chromatin structure then allows for the recruitment of the transcriptional machinery. This delicate balance of acetylation and deacetylation is reversibly catalyzed by multiple classes of histone acetyl transferases (HATs), which add acetyl groups to histone lysine residues, and HDACs, which can remove these modifications [42–45].

HDAC Inhibitors

A wide range of naturally occurring and synthetic compounds have been found to inhibit the activity of class I and class II HDACs, some of which are now Food and Drug Administration-approved and in use for their effectiveness as anticancer or neuropsychiatric agents [46, 47]. Inhibitors of the class I and II HDACs are categorized into several different classes, based on their chemical structure [47]. HDAC inhibitors (HDACi) can be divided into structural classes, including 1) small-molecule hydroxamates, such as Trichostatin A (TSA), suberoylanilide hydroxamic acid (SAHA), scriptaid, and oxamflatin; 2) short-chain fatty acids, such as sodium butyrate, sodium phenylbutyrate, and valproic acid (VPA); 3) cyclic tetrapeptides, such as apicidin, trapaxin, and the depsipeptide FK-228; and 4) benzamides, such as MS-275 and Cl-994 (for a review see [48]).

Young and Old Animals and Cells are Very Different in Their Response to Lesion

Some of the most promising strategies to promote motor function following brain injury, stroke, and SCI in humans and experimental animals involves targeted training and rehabilitation to enhance residual plasticity. Such approaches are a constantly evolving “state of the art” treatment approach, but successful rehabilitation outcomes vary significantly and depend on several variables, including the health and emotional motivation of the individual. Perhaps the variable producing the greatest shift in rehabilitation-based recovery is age [49–51], which is also a critical feature of likelihood of morbidity within the first year after SCI [52]. Whether in patients or animal models, the state of developmental plasticity of the nervous system and its accompanying immune responses clearly underscores its fundamental capacity to drive functional recovery [53–55]. Age at injury is a prognostic indicator in animal models regardless of whether locomotor recovery is spontaneous [56] or stimulated by training [57, 58]. During recovery from lesion, do younger rats and humans remain more plastic because they are less restricted (fixed) epigenetically? If so, understanding the epigenetic basis underlying the differential plasticity of “young” and “old” cells could open avenues that allow us to specifically target chromatin-based plasticity at key stages of the recovery process to enhance rehabilitation-stimulated functional recovery.

Although chromatin-remodeling activities are integral to the proper development of the nervous system, the potential effect of altering such activities in adult brain repair is only just beginning to be explored, and its success will rely on understanding the chromatin remodeling factors that are specific to distinct developmental time windows in the cells that need to be most manipulated following SCI.

Epigenetic Regulation of Nervous System Developmental Plasticity

Epigenetic mechanisms can alter distinct cell gene expression profiles in vivo without directly affecting DNA sequence, and thus influence transcription in a malleable manner with far-reaching implications for understanding individual human diversity, and variability in injury and disease responses [59–61]. No other tissue or organ rivals the brain’s ability to diversify transcriptional programs, enabling complex patterns of postmitotic gene regulation [27]. The progressive differentiation of individual neurons and glial cells is driven by a fine balance of transcriptional activation, and the epigenetic regulation of chromatin structure, with histones, in particular [62], increasingly implicated in modifying neuronal learning, synaptic plasticity, and cognition [26]. Our understanding of how some chromatin remodeling factors function in the brain has been illuminated by the multiple neurological syndromes in which they are mutated. This includes Rett syndrome, (Immunodeficiency–centromeric instability–facial anomalies) ICF, fragile-X syndrome, ATRX (Alpha-Thalassemia X-Linked Intellectual Disability), Rubinstein–Taybi, and Angelman syndrome (reviewed in [59]), and complex psychiatric disorders such as schizophrenia [63–65]. Similarly, a large number of neural-related phenotypes are beginning to emerge from genetic studies of chromatin remodeling proteins in animal models (see [66] for a review).

It is not hard to imagine that holding in place the remarkably complex patterning of the brain, whilst supporting the stimulus-dependent plasticity that allows the brain to “learn”, may require higher-order gene regulation mechanisms in order to distinguish billions of otherwise genomically-identical neurons from each other. In accord, DNA methylation of distinct genomic sites in the human brain increases with activity and neuronal learning [67], producing a distinct “methylation signature” in different adult human brain regions [68]. Many questions remain, however, concerning how life experience affects other chromatin remodeling factors in individual cells of the adult CNS. A large variety of environmental stimuli (e.g., stress, nutrition, nurturing, exposure to drugs of abuse, social interactions, sensory learning, etc.) can alter the expression and activity of DNA methylation and histone-modifying enzymes throughout the neuro-genome [66, 69, 70]. Epigenetic regulation also provides a coordinated system for regulating gene expression at each stage of neurogenesis—where each neuron assumes a distinct chromatin profile as a result of environmental stimulation over the lifespan [25, 71, 72]. Collectively, less mature (“plastic”) cells should be able to more rapidly change state and respond to the needs of an injured environment better than older, more specialized, or senescent cells. Following CNS injury, it may thus be advantageous to help cells revert to a more plastic (“open chromatin”) genomic state after injury, so they can adapt as they once did developmentally, to promote the restoration of function. This may be even more critical to consider in approaches that incorporate neural stem cell-based approaches to promote regeneration.

Could Altering Cellular Epigenetic Changes in State Enhance SCI Repair?

So, epigenetic mechanisms are clearly essential for normal brain development, and chromatin remodeling is now linked to neural plasticity at the molecular, cellular, and organismal level, and multiple forms of behavioral learning [61, 73–75]. However, promoting long-term neurological recovery—in SCI and TBI in particular—will not only require us to promote learning and plasticity in neurons, but also manipulate the orchestrated responses of glia, de novo angiogenesis, and central and peripheral immune cells. The next frontier of using epigenetics to enhance SCI repair will depend on understanding how distinct chromatin modifications drive function in individual cells that contribute to SCI pathophysiology (Fig. 1). Understanding the correlation between chromatin state and immature (plastic, developing) or differentiated (committed) function in non-neuronal cells will be an essential part of minimizing long-term damage and supporting functional recovery following SCI. Below, we first introduce how environmental factors (like nutrition) may create epigenetic changes and review known pathways activated after SCI that have recently been shown to affect chromatin remodeling directly. We next discuss how histone acetylation status (which is reversible) is mediated by different enzymes in different cell types, and how this could be potentially targeted to enhance endogenous repair of the spinal cord.

Some Established Pathways Activated After SCI Alter Chromatin Remodeling

Interestingly, some of the major signaling cascades that are differentially activated in the CNS in development and after injury have recently been implicated in controlling histone acetylation and chromatin structure. First, the mitogen-activated protein kinase (MAPK) superfamily is central to plasticity signaling in the CNS. Its prototype is the extracellular signal-regulated kinase (ERK)/mitogen- and stress-activated kinase (MSK)/cyclic-adenosine monophosphate regulatory element binding protein (CREB) pathway, where ERK activates its downstream target, MSK, to phosphorylate CREB [76–79]. This phosphorylation and activation of CREB recruits CREB binding protein, a HAT that regulates local chromatin structure as part of CREB-dependent activation of nuclear gene transcription [80]. Second, the nuclear factor kappa B (NFκB) signaling pathway appears to control histone acetylation and chromatin structure in the CNS by mechanisms that are still being elucidated [81, 82]. In the immune system and CNS, NFκB is controlled by its upstream regulator, inhibitor of kappa B kinase, which itself is a target of multiple upstream regulatory signaling cascades affected by chromatin remodeling factors. Thus, both the ERK/MSK/CREB pathway and the inhibitor of kappa B kinase/NFκB pathway can dynamically regulate chromatin structure in the mature CNS. In addition, some class II HDACs can act on nonhistone proteins like the transcriptional co-repressor YY1, the neuronal protein alpha-tubulin [83], and NFκB itself [84–87] to alter their function. Finally, peroxisome proliferator-activated receptor gamma (PPARγ) function is dependent on the availability of co-regulator proteins that modify chromatin states. As a result, chromatin modifying factors could differentially regulate the transcriptional activities of PPARγ and its target genes. Many known co-activators and co-repressors of PPARγ and other nuclear receptors have intrinsic histone modifying activities [88].

Nutritional Changes Can Affect Chromatin Structure and SCI Recovery

Changes in prenatal and perinatal nutrition can modify epigenetic state, thereby altering gene expression, neurogenesis, and behavior in offspring (reviewed in [89–91]). The nutrients that appear to have the greatest effects on brain plasticity include protein, folate, iron, selenium, zinc, iodine, vitamin A, and docosahexaenoic acid. Mechanistically, protein-restricted diets can inhibit DNMTs and cause hypomethylation of specific gene promoters [92], an upstream event of HDAC recruitment and activation. In addition, folate and vitamin B12 are essential cofactors for the methylation cycle; thus, deficiencies in these vitamins also inhibit DNMTs and DNA methylation, which can cause oxidative stress and neuronal cell death [93–96]. Whereas nutrition and health status have long been debated for their contribution to SCI recovery, nutrition and metabolism are now implicated bi-directionally in determining epigenetic status—with different potential effects across individuals [97]. This might explain, in part, studies where manipulating timing and type of caloric intake can enhance some parameters of recovery following SCI [98, 99].

Cells Regulated by HDACs That Could be Targeted to Alter the Course of Secondary Damage in SCI

Chromatin maintenance in dividing and nondividing cells relies on the action of replication-independent histone variants. Even simply understanding variability in distribution of histone variants could greatly affect our understanding of the plasticity of individual cells, or even reveal mechanisms underlying diverse repair responses between individuals. A failure to regenerate after SCI has 3 major cell groups that could each serve as independent targets for histone manipulation to enhance endogenous repair: 1) neurons with different intrinsic capacities to initiate a regeneration or plasticity response; 2) a hostile glial and immune response that is inhibitory or detrimental to repair; and 3) endogenous progenitors that could be activated to replace lost cells or promote remyelination.

Chromatin and Neuronal Regeneration

Chromatin-remodeling has now emerged as a core mechanism mediating neuronal differentiation and activity-dependent transcription, a process important for plasticity in both development and adulthood (see [66, 100, 101]). However, in order to regenerate, a neuron must first survive, and maintaining the balance between the activity of HATs and HDACs also appears to be pivotal for neuronal survival, even in the absence of pathology [102, 103]. Direct effects of HDACi (and other alterations in chromatin structure) on neuronal survival are covered in more detail elsewhere in this issue. Here, we focus on how chromatin modification may enhance neuronal outgrowth following injury—an essential step in rebuilding linear or alternative circuitry in a superhighway like the spinal cord.

Developing a functional circuit requires 2 major components—the structural components that build axons and dendrites and their transport systems, and the molecular pathways that direct axons and match them to their appropriate targets [104]. During CNS development, oriented axonal growth results from combinations of factors that attract or repulse the movement of neuronal processes [105, 106] (reviewed in [104, 105, 107, 108]). Forces driving neurite growth are intrinsic to developing neurons (motor machinery, cytoskeleton, metabolism, surface receptors), whereas guidance cues are programmed by the environment. Once targets have been reached, both of these sets of machinery are generally suppressed (some by chromatin-mediated silencing) in order to allow that neuron to mature in response to stimulation. Although many of these events may need to be reinitiated to promote regeneration, most are no longer available in the complex milieu of the lesioned, mature nervous system [109–112].

Chromatin remodeling factors are now emerging as regulators of chromatin remodeling that may gate access to genes essential for axon and dendrite growth and regeneration. Neurological disorders involving chromatin modification suggest that epigenetic silencing of transcription for outgrowth in mammalian CNS neurons may help to explain the developmental loss of intrinsic neurite growth and targeting capacity upon maturation [113]. For example, GAP43—a prototypical “regeneration-associated gene” is directly silenced by MeCP2, a methyl–DNA binding protein induced in terminally differentiating neurons that recruits HDAC2 to silence developmental genes [114]. When mutated, MeCP2 results in Rett syndrome, where neurons retain a prolonged immature state [115]. In accord with this, histone acetylation levels (histone 3 (H3) K9/14) increase in cortical neurons and cerebellar granule neurons as they mature [116]. The expression of epigenetic regulators also appears to tip the balance of attenuating transcription following stroke in cortical neurons, where neurons that do not sprout up-regulate HDAC4 and down-regulate p300 (HAT) compared with their axon-sprouting counterparts [117]. Furthermore, TSA can increase neurite growth in postnatal day 7 cerebellar granule neurons on permissive and inhibitory substrate, and enhance GAP43 expression and H3K9/14 hyper-acetylation [116].

However, TSA can also stimulate neurite growth beyond the nucleus by enhancing acetylation of transcription factors or proteins like α-tubulin that are substrates for axonal elongation [118–121]. HDAC6 also functions largely outside the nucleus, with a number of cytosolic and axonal targets that could affect endogenous outgrowth (reviewed in [122]). HDAC1 can also be shuttled out of the nucleus and retained more highly in the cytoplasm in damaged neurons in response to inflammatory stimuli. Once in the cytosol, it is directed to axons where it can impair mitochondrial transport and induce neurite beading in response to excitatory amino acids and cytokines [120]. Cytosolic HDAC1 has been detected in damaged axons in brains of multiple sclerosis patients, in animal models of demyelination (cuprizone), and in cultured neurons exposed to excitatory amino acids (glutamate) and cytokines [tumor necrosis factor (TNF)-α]. Although cytoplasmic acetylation targets (like tubulin) could alter neurite outgrowth, a number of lines of evidence in cortical neurons [117], cerebellar granule neurons, and retinal ganglion cells [116] indicate that HATs, CREB binding protein/p300, and P/CAF (P300/CBP-associated factor) are required for their hyperacetylation-induced increase in neurite growth on inhibitory substrates. However, p300 and HDAC inhibition may act on other independent aspects of neuronal outgrowth (survival, then outgrowth signaling) and may need to be considered independently for the more critical activity following a severe injury like SCI, especially when considering the complexity of the injury environment compared to an in vitro assay. Therefore, if HDAC inhibition is to be used to promote neuronal survival and regeneration following SCI in vivo, it could be important to target its use to the most beneficial acute time window, provided the glial-immune environment can support regrowth.

Can HDAC Manipulation Help to Break Down Glial–immune Barriers?

Following focal stroke, TBI, and SCI, a series of immediate changes in inflammatory signaling that directly result from the extent and type of impact initiate the second, prolonged phase of subacute inflammatory responses [123] (Fig. 1). Macrophages play a major role in both the acute and sustained SCI response, with 2 major macrophage phenotypes (M1 and M2) emerging as playing different roles that shift the balance of neurotoxicity and regeneration in the injured mouse spinal cord [124]. M1 macrophages, stimulated by interferon-gamma (IFNγ), are the “classical” pro-inflammatory macrophage subtype. They express CD86, inducible nitric oxide synthase (iNOS), CD16/32, and major histocompatibility complex II, and produce high levels of oxidative metabolites and pro-inflammatory cytokines. Although essential for host defense and tumor cell killing, they can cause collateral damage to healthy cells/tissue, especially in a “closed” system like the CNS. M2 macrophages express CD206 and Arginase 1 in presence of interleukin (IL)-4 or IL-13, and can enhance angiogenesis, suppress destructive immunity, decrease lesion size, and also promote long-distance axon growth of dorsal root ganglion neurons—even in the presence of growth inhibitory substrates [124]. M1 and M2 are both activated following SCI, but M2 declines and M1 persists (after the first week), resulting in prolonged inflammation. The M2 state can be maintained by PPARγ agonists, which are neuroprotective in SCI. Epigenetic status appears key in regulating the phenotype of macrophages and T-cells during their responses to injury, and HDAC inhibition could promote repair, whilst minimizing inflammatory damage [125]. IFNγ (from activated T-cells) and Toll-like receptor activation can enhance histone acetylation in macrophages, and promotes the M1 phenotype. HDAC3 is a direct regulator of transcription factors promoting M1 in the alternative activation of M1 and M2 states in response to cytokines like IL-4. When HDAC3 is deleted, it allows for the accumulation of M2 macrophages, which produce a pro-repair, anti-inflammatory environment [126]. This balance of reactions involves both PPARγ and Jmjd3m, a demethylase that is critical for M2 polarization in response to concurrent activation by NFκB and Toll-like receptor [127].

Following SCI, microglia also undergo a distinct series of morphological and biochemical changes in state that can be distinct from those of microglia in the brain following physical injury, and can directly lead to acute and prolonged neuropathic pain [128]. Following injury, spinal cord microglia rapidly up-regulate TNF-α, IL-1β, IL-6, nitric oxide (NO), Bradykinin (BK), matrix metalloproteinases, and cathepsin S, all of which initiate and maintain neuropathic pain. In particular, microglial cathepsin S contributes to acute pain following injury via the p38 MAPK pathway—a reaction that can be inhibited by the broad-spectrum HDACi, VPA. Some of these same signaling changes occur in microglia after TBI, where microglia and macrophages become hypomethylated concomitant with a shift in activation state [129]. DNA methylation and histone deactylation changes may thus determine the strength and extent of the immune response following CNS lesion. A different HDAC-specific inhibitor, MS-275 (which preferentially targets HDAC1 and 2), applied 10 days after the initiation of rat experimental autoimmune neuritis suppressed influx of T-cells, B-cells, and macrophages, and facilitated recovery, at the same time as minimizing long-term damage in the rat. MS-275 can also enhance the production of M2 macrophages over M1 [130].

Astrocytes undergo reactive astrogliosis in response to all forms of CNS insults—infection, trauma, ischemia, and neurodegenerative disease—by undergoing hypertrophy and changing gene expression. In rodent models of SCI, astrocytes can be either beneficial or detrimental after injury, depending on the type of lesion and the immediate inflammatory environment. Many of the cytokines released after SCI stimulate reactive astrogliosis, glial scar formation, and neurotoxicity, and can impede endogenous repair (reviewed elsewhere [131–133]). During reactive gliosis there are increases in glial fibrillary acidic protein (GFAP) and cyclooxygenase (COX)-2 in astrocytes, and increases in iNOS, NO, IL-6, and TNF-α in both astrocytes and microglia (reviewed in [122]). Each of these genes can be regulated by HDAC inhibition—usually to attenuate the inflammatory response— in different lesion scenarios. In addition, reactive astrocytes up-regulate glycosaminoglycans, like chondroitin sulfate proteoglycan, and some glycosaminoglycans (e.g., heparin) can themselves act as HAT inhibitors and oppose potential increases in acetylation levels in other cells [134]. It is thus challenging to analyze the effect of HDAC and HAT activation independently on astrocytes following SCI because their activity is intrinsically related to the proximity and state of local microglia and macrophages, the cytokines of which can alter astrocyte acetylation.

Multiple lines of evidence indicate that HDACs play a core role in regulating astrocytic gene expression. Differentiation of neural stem cells into astrocytes appears to require the recruitment of the co-activator complex members, including STAT (Signal Transducer and Activator of Transcription) 1/3, HATs and Smad1 to specific regions of the Gfap promoter, where GFAP is the “prototypic” astrocyte gene [135]. Treatment with HDACi increases the expression of S100 and Gfap in cultured oligodendrocyte cells (OPCs) in neonatal rats [136]. Furthermore, Gfap is actively repressed by complexes containing HDAC, the nuclear receptor co-repressor and adaptor proteins [137], where nuclear receptor co-repressor knockout mice are characterized by precocious expression of Gfap [137, 138].

HDACi and Astrocytes

The responses of astrocytes and microglia to HDAC inhibition is highly context-dependent, and can change with species, HDACi used, and assay system employed. In human microglia and astrocytes in vitro, for example, HDAC inhibition with TSA or VPA can suppress cytokine and chemokine gene expression, with different effects on different groups of cytokines in each cell type [139]. An alternative HDACi, SAHA, also suppresses IFNγ-induced neurotoxicity of human astrocytes through inhibition of the STAT3 signaling pathway. Sodium butyrate and TSA both induce COX-1 (but not COX-2) in a normal astrocyte cell line [140], even though COX-2 expression can be regulated by HDAC inhibition in other non-neural cells (see [141]). Similarly, in mixed cultures of mouse astrocytes and microglia, suppression of HDAC activity using SAHA and ITF2357 (a SAHA analog), inhibited the inflammatory response to lipopolysaccharide (LPS) by direct impairment of transcriptional machinery, with a dramatic inhibition of iNOS and COX-2 induction [142]. In mouse cells, sodium butyrate also inhibits hypoxia-induced iNOS protein and down-regulation of TNF-α messenger RNA, thus reducing the inflammatory response [143]. Sodium butyrate can also induce an adaptive response to LPS-stimulated microglial activation by attenuating NO, IL-6, and TNF secretion—all classic microglial inflammatory responses [144]. However, this anti-inflammatory activity may be specific to short-chain fatty acid HDACi, as the hydroxamate HDACi, TSA and SAHA, strongly potentiate the LPS-induced inflammatory response in primary microglial, neural co-cultures, and hippocampal slices [145]. Activated microglia-conditioned medium can induce HDAC activity in astrocytes (through p38 MAPK or glycogen synthase kinase 3 beta (GSK3β) activation), decreasing H3 acetylation, down-regulating Nrf2 (astroglial nuclear factor involved in anti-oxidant defense), thus decreasing protection. Inhibitors of HDACs, p38 MAPK, and GSK3β, can ameliorate this response in vitro, thus setting the stage for assisting in neuroprotective mechanisms in vivo [146].

Through either direct modulation of transcriptional activity or inhibition of inflammatory signaling in microglia and astrocytes, HDACi’s could, ostensibly, decrease chemokine and cytokine release, decrease neurotoxicity, enhance protection against anti-oxidative damage, and promote functional recovery in rat models of SCI. Thus, modulation of HAT and HDAC function through targeted inhibition could powerfully modulate astrocyte responses, enhance protection, and reduce barriers to regeneration after lesion [133].

Histone Modifications Drive Transitions from Embryonic Stem Cells to Multipotent Neural Stem Cells to CNS Precursors and Differentiated Cells

Chromatin remodeling plays a distinct role during progressive restriction of cell lineage in CNS differentiation at 3 stages (reviewed in [25]): 1) the transition from pluripotent embryonic stem (ES) cells to multipotent neural precursors—where the ES cell chromatin state is comparatively “open” in order to generate all the cell types required in an organism, and suppress expression of genes for lineage commitment and differentiation, whilst retaining DNA in a conformation that is poised for transcription [147]; 2) the restriction of multipotent neural precursors to lineage committed neural or glial precursors—characterized by the stable silencing of genes involved in other lineages (i.e., endoderm and mesoderm) and expression of neural precursor genes; 3) terminal differentiation of specialized cells (i.e., neurons, astrocytes, and oligodendrocytes) from precursors [148]. In considering how to manipulate chromatin to better stimulate adult neural stem cells (NSC) to participate in endogenous repair, or to control stem cells prior to or after SCI transplantation, it is important to understand how the chromatin state of NSCs in adult differs from those in the embryo [149]. This may be particularly important in the spinal cord, where our understanding of adult progenitor subtypes lags far behind that in the adult brain [150].

Histone Acetylation Alters Neural Stem Cell Function

HDACs 1 and 2 are clearly critical for CNS development as their ablation results in disorganization of brain structures and postnatal lethality [151]. In accord, during NSC differentiation, HAT p300 is recruited to GFAP during astrogliogenesis, controlling its expression [152, 153]. Removal of HDAC complexes from neurogenic transcription factors such as NeuroD is also necessary for neuronal differentiation to proceed [154]. The action of HDACi on different classes of CNS progenitors is context- and age-dependent. For example, both VPA and TSA can revert committed oligodendrocyte precursors to multipotent neural progenitors [136, 155], and HDAC inhibition can favor neuronal differentiation at the expense of glial differentiation in both embryonic and adult progenitors in vitro [154, 156, 157]. Furthermore, astrocytes and oligodendrocytes differentiated in vitro from adult hippocampal progenitors have lower levels of acetylated histones H3 and H4 than undifferentiated progenitors or neurons, suggesting that there is more deacetylation during the differentiation of glial lineages [154].

VPA may thus increase neuronal differentiation from adult neural progenitor cells whilst inhibiting astrocyte and oligodendrocyte differentiation [154]. Treatment of rats with VPA during the first 2 weeks postnatally induces a stalling of oligodendrocyte differentiation in the developing rat corpus callosum, which is reversible if VPA is removed for 2 days [158]. We have also recently shown that postnatal Subventricular zone neural stem cell activity and inhibitory neurogenesis, in particular, is also profoundly altered in vivo and in vitro by short-term treatment with both VPA and TSA [72]. In contrast, treatment with VPA after the third week postnatally, when differentiation is largely complete, has little effect on the expression of oligodendrocyte stage-specific proteins, or myelination, indicating a critical time window and transitional function for histone deacetylation in the differentiation of oligodendrocytes [158]. Neuronal differentiation may therefore be a default pathway in the absence of specific HDAC activity, which serves to permit glial differentiation in key developmental time windows.

Remyelination: Epigenetic Regulation of Oligodendrocytes

Oligodendrocytes play 2 opposing roles in a potential repair scenario following SCI. First, their myelin components play a profoundly inhibitory role to axonal regeneration [159]. Even lesioned spinal axons that do have the capacity to initiate a regeneration program may be stopped in their tracks by mature oligodendrocytes and their myelin proteins [160]. Holding back the differentiation of oligodendrocytes, whilst clearing myelin debris, will likely enhance endogenous sprouting and regeneration in spinal pathways. Furthermore, spared axons in the peri-lesion areas become rapidly demyelinated after lesion, and must be re-myelinated in order to restore function. Understanding HDAC action within OPCs mechanistically will be a critical component of identifying time windows after SCI in which to first inhibit OPC differentiation during a period of neurite outgrowth, then allow HDAC action in order to restore remyelination programs. A series of elegant studies has clearly demonstrated that histone deacetylation is critical for oligodendrocyte differentiation (reviewed in [161]), and should be carefully considered in the manipulation of chromatin plasticity in a SCI scenario. Coupled with the profound effect that HDAC inhibition could have on producing myelin to support excitatory neurotransmission [162], HDAC inhibition during repair could result in long-term beneficial shifts in the excitatory/inhibitory balance in different neuronal circuits, depending on the critical period of treatment [163].

HDACi in SCI Repair—The Story So Far

HDAC inhibition is also now being explored clinically to treat neuropsychiatric disease and neurodegenerative disorders [164], but only limited studies have directly tested HDAC inhibition in SCI models. When delivered by minipump in vivo, VPA can attenuate microgliosis in lesioned spinal cord, and purinergic P2X₄R expression in activated microglia, which is associated with neuropathic pain [165]. VPA treatment in vitro appears to decrease microglial activation, whereas, in contrast, TSA and sodium butyrate appear to enhance activation. VPA treatment can also enhance neuronal protection and improves open-field behavioral assays following SCI [165]. In an alternative SCI model, VPA also reduced cavitation and gliosis, enhanced neuronal sprouting, and increased the endogenous production of brain- and glial-derived neurotrophic factors around the lesion site [166]. These behavioral or pathological improvements are suggested to be due, in part, to changes in H3 and H4 histone acetylation as early as 1 day after treatment [167]. In addition, VPA concurrent with transplantation of NSCs also enhanced their survival, integration and migration, and improved their neuronal differentiation, promoting some functional recovery [168]. Despite variations in these findings—perhaps owing to varying methods of administration (intravenous, intrathecally, or by mini-pump)—VPA could be effective in SCI if given early after injury, but has many off-target effects that may make it less desirable. Finally, drugs that target class II (but not class I) HDACs have recently emerged as potential tools for inhibiting the development of inflammatory hyperalgesia [169], a common problem after SCI. Thus, cell-specific action of distinct HDACs appears to be the beginning of being able to specifically target the cells and symptoms that accompany different phases of SCI (summarized in Fig. 2).

Fig. 2.

Epigenetic regulation of histone acetylation in different cells participating in spinal cord injury repair. Acetylation of lysine residues in histone tails constitutes the major post-translational histone modification. Histone Acetyl Transferase (HAT) and histone deacetylase (HDAC) can affect many cell functions through chromatin remodeling or post-translational modifications of transcription factors and cytoplasmic proteins. HDAC inhibitors (HDACi) act on many cell types in the central nervous system: neurons, astrocytes, oligodendrocytes, microglia, and neural progenitors. OPC = oligodendrocyte precursor cells

Moving Towards Targeted Chromatin-Based Therapy for SCI

Unlocking the activity of pro-regenerative transcription factors and allowing greater access—and recruitment [30]—to promoters of pro-regenerative genes could be a powerful approach to drive neuronal regeneration. Equally important, manipulating chromatin to block access to gene loci in glia and inflammatory cells that currently inhibit regeneration could transiently remove blocks to repair. In an ideal SCI repair scenario, you would want—in defined time windows—to 1) minimize neuroinflammation and promote protection; 2) promote regeneration of spared neurons; 3) harness progenitor activity to best respond to the needs of the environment; and 4) promote re-myelination of spared and regenerated tracts. Each of these processes can be differentially impacted by HDAC inhibition—depending on the experimental paradigm used. Timing truly is everything if we are to consider targeting individual cell types at different phases of SCI pathology to promote repair, at the same time as minimizing side-effects (Figs. 1 and 2).

A number of HDACi’s are already in use as anticancer agents because they are able to increase genes involved in growth arrest and promote apoptosis of cancer cells, but new-generation and highly-specific HDACi are emerging almost weekly. Use of HDACi for therapeutic purposes within the CNS is, however, still a topic of intensive debate, as their mechanisms of action in different CNS cells are only beginning to be understood, and histone manipulation yields conflicting results at different ages, even in the same model [170]. We are approaching a time when targeting HDAC inhibition in time and space and carefully monitoring its activity will be possible. New inhibitors aimed at individual HDACs enriched in distinct cells, coupled with assays of acetylation and methylation, and high-throughput genomic analysis of chromatin modification in small numbers of peripheral and CNS cells will allow us to test the specificity of response. Luciferase reporter mice allow us to observe the efficacy of intervention during the lesion response in vivo. Biomarkers analysis from serum and cerebrospinal fluid that change with glial-immune activation will also help to better monitor the effectiveness of this kind of pharmacological intervention.

With clinical data already highlighting a broad range of pathological and functional outcomes in human SCI patients with the same lesions, we should begin to consider the effect that an individual’s life experience has had on their epigenome, and better assess how to build upon this to enhance and design their own profile of recovery. At the brink of a revolution in personalized medicine, fully understanding a patient’s epigenetic status at the time of injury may not only help us to understand disparities in outcome, but also help to define the best course of action in manipulating their own epigenetic status to facilitate optimal functional recovery.